hdx_gpt-3_tag_prediction.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Introduction\n",
    "\n",
    "The [Humanitarian Data Exhcange](https://data.humdata.org/) (HDX) is a fantastic platform for storing humanitarian datasets, key to timely and efficient response to humanitarian disasters. The data on the this platform is sourced from all around the world. One key aspect to this platform is the use of [Humanitarian Exchange Language](https://hxlstandard.org/?_gl=1*p7ffaf*_ga*OTc2MzE0MTI5LjE2NzAxMDE2NjE.*_ga_E60ZNX2F68*MTY3MzQ5NDk2NS4xOC4xLjE2NzM0OTUzNjQuNjAuMC4w) (HXL) tags. This meta data provides important meta data enables faster automatic integration and interoperability, as well as a way to link datasets together. \n",
    "\n",
    "Tags come in two forms, those set at the dataset level, and field-level tags which apply to columns in tabular data, which look like:\n",
    "\n",
    "\n",
    "\n",
    "| FDF       | PAHO Sub-Region     | ISO-3              | Country / territory | IFRC Sub-Region | Population        | Tracker Update Date  | Data Publication Date  | EPI Week (year)    | EPI Week (outbreak)    | Cases     |\n",
    "|-----------|---------------------|--------------------|---------------------|-----------------|-------------------|----------------------|------------------------|--------------------|------------------------|-----------|\n",
    "| #severity | #region             | #country+code+iso3 | #country+name       | #region+ifrc    | #population+total | #date+tracker+update | #date+data+publication | #date+epiweek+year | #date+epiweek+outbreak | #affected |\n",
    "| Confirmed | Non Latin Caribbean | BRB                | Barbados            | Americas        | 283380            | 15th February 2016   | 2016-02-12             | 6                  | 21                     | 7         |\n",
    "| Deaths    | Non Latin Caribbean | BRB                | Barbados            | Americas        | 283380            | 15th February 2016   | 2016-02-12             | 6                  | 21                     | 0         |\n",
    "| Suspected | Non Latin Caribbean | BRB                | Barbados            | Americas        | 283380            | 15th February 2016   | 2016-02-12             | 6                  | 21                     | 0         |\n",
    "\n",
    "[ *You can see more examples [here](https://hxlstandard.org/hxlexample/)* ]\n",
    "\n",
    "Notice the second row just below the column headers, those are the tags. They consist of the 'Tag' which is prefixed with a `#` and in some cases following attributes, eg `#country+code+iso3`\n",
    "\n",
    "The issue is that these field-level tags are not set on all datasets in HDX, making it harder to use data. There has been some work previously on predicting HDX field tags, see [this notebook](https://github.com/humanitarian-data-collaboration/hdx-python-model) and [this paper](https://www.kdd.org/kdd2019/docs/Humanitarian_Data_tagging_KDD2019_SocialImpactTrack_HXLTagPrediction.pdf), but this work predicts just the tag and not the attributes after the tag. \n",
    "\n",
    "This notebook demonstrates how we can use [GPT-3](https://openai.com/blog/gpt-3-apps/) to predict not only HXL field tags, but also tag attributes. \n",
    "\n",
    "Matthew Harris\n",
    "\n",
    "# Setup\n",
    "\n",
    "1. See Conda environment `./environment.yml` if you want to run this notebook locally\n",
    "2. For running OpenAI you will need to create a file called `api_key.txt` in the current directory and put your [OpenAI API key](https://beta.openai.com/account/api-keys) in there (just the API string, nothing else)\n",
    "3. Edit the variable 'country' below to set which country you wish to prredict tags for\n",
    "\n",
    "No setup is needed for HDX integration as we're using open datasets and reading, so they are open in the API."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: html5lib in /Users/matthewharris/opt/miniconda3/envs/dk_analysis/lib/python3.8/site-packages (1.1)\n",
      "Requirement already satisfied: six>=1.9 in /Users/matthewharris/opt/miniconda3/envs/dk_analysis/lib/python3.8/site-packages (from html5lib) (1.16.0)\n",
      "Requirement already satisfied: webencodings in /Users/matthewharris/opt/miniconda3/envs/dk_analysis/lib/python3.8/site-packages (from html5lib) (0.5.1)\n",
      "Note: you may need to restart the kernel to use updated packages.\n"
     ]
    }
   ],
   "source": [
    "# Installation of the api\n",
    "#%pip install hdx-python-api==5.9.3\n",
    "#%pip install sklearn==1.2.0\n",
    "#%pip install openai==0.26.0#\n",
    "#%pip install bs4==0.0.1\n",
    "#%pip install lxml==4.9.2\n",
    "#%pip install html5lib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 324,
   "metadata": {},
   "outputs": [],
   "source": [
    "from hdx.utilities.easy_logging import setup_logging\n",
    "from hdx.api.configuration import Configuration\n",
    "from hdx.data.dataset import Dataset\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import pandas as pd\n",
    "import json\n",
    "import os\n",
    "import chardet\n",
    "import shutil\n",
    "import sys\n",
    "import re\n",
    "import traceback\n",
    "import zipfile\n",
    "import time\n",
    "\n",
    "import numpy as np\n",
    "from sklearn.model_selection import train_test_split\n",
    "import openai\n",
    "from openai import cli\n",
    "\n",
    "from sklearn.metrics import precision_score, recall_score, f1_score, accuracy_score\n",
    "\n",
    "import random\n",
    "random.seed(42)\n",
    "\n",
    "# Open AI API key should be put into this file\n",
    "openai.api_key_path = \"./api_key.txt\"\n",
    "\n",
    "pd.set_option(\"display.max_rows\", 500)\n",
    "\n",
    "country = \"Kenya\"\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Get HDX Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\u001b[32m2023-01-12 17:42:21.522\u001b[0m | \u001b[1mINFO    \u001b[0m | \u001b[36mhdx.api.configuration\u001b[0m:\u001b[36m__init__\u001b[0m:\u001b[36m95\u001b[0m - \u001b[1mNo HDX base configuration parameter. Using default base configuration file: /Users/matthewharris/opt/miniconda3/envs/dk_analysis/lib/python3.8/site-packages/hdx/api/hdx_base_configuration.yml.\u001b[0m\n",
      "\u001b[32m2023-01-12 17:42:21.523\u001b[0m | \u001b[1mINFO    \u001b[0m | \u001b[36mhdx.api.configuration\u001b[0m:\u001b[36m__init__\u001b[0m:\u001b[36m98\u001b[0m - \u001b[1mLoading HDX base configuration from: /Users/matthewharris/opt/miniconda3/envs/dk_analysis/lib/python3.8/site-packages/hdx/api/hdx_base_configuration.yml\u001b[0m\n",
      "\u001b[32m2023-01-12 17:42:21.536\u001b[0m | \u001b[1mINFO    \u001b[0m | \u001b[36mhdx.api.configuration\u001b[0m:\u001b[36m__init__\u001b[0m:\u001b[36m133\u001b[0m - \u001b[1mNo HDX configuration parameter and no configuration file at default path: /Users/matthewharris/.hdx_configuration.yml.\u001b[0m\n",
      "\u001b[32m2023-01-12 17:42:21.537\u001b[0m | \u001b[1mINFO    \u001b[0m | \u001b[36mhdx.api.configuration\u001b[0m:\u001b[36m__init__\u001b[0m:\u001b[36m222\u001b[0m - \u001b[1mRead only access to HDX: True\u001b[0m\n"
     ]
    }
   ],
   "source": [
    "# Logging in and getting read only access to https://data.humdata.org\n",
    "# Read more about API Access token in documentation at https://hdx-python-api.readthedocs.io/en/latest/\n",
    "# for writing access.\n",
    "\n",
    "agent_count = 0\n",
    "\n",
    "def setup_hdx_connection(agent_name):\n",
    "    setup_logging()\n",
    "    Configuration.create(hdx_site=\"prod\", user_agent=agent_name, hdx_read_only=True)\n",
    "\n",
    "\n",
    "setup_hdx_connection(f\"DK_UserAgent{agent_count}\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Build a master list of datasets, resources, tags and groups for all countries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_json_from_hdx_meta(hdx_object):\n",
    "    \"\"\"\n",
    "    This function takes a HDX Python object as input and returns a JSON object.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    hdx_object : HDX object\n",
    "        The HDX object to be converted to JSON.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    json_op : JSON object\n",
    "        The JSON object created from the HDX object.\n",
    "    \"\"\"\n",
    "    json_op = {}\n",
    "    for key in hdx_object:\n",
    "        json_op[key] = hdx_object[key]\n",
    "    json_op = json.dumps(json_op, indent=4)\n",
    "    return json.loads(json_op)\n",
    "\n",
    "\n",
    "print(f\"Searching for ALL datasets in HDX\")\n",
    "datasets = Dataset.search_in_hdx()\n",
    "\n",
    "print(f\"Extracting resource list for each dataset\")\n",
    "resources_output = []\n",
    "tags_output = []\n",
    "groups_output = []\n",
    "\n",
    "for d in datasets:\n",
    "    try:\n",
    "        dataset = Dataset.read_from_hdx(d[\"id\"])\n",
    "    except:\n",
    "        print(f\"Error reading dataset: {d['id']}\")\n",
    "        time.sleep(30)\n",
    "        dataset = Dataset.read_from_hdx(d[\"id\"])\n",
    "\n",
    "    for tag in dataset[\"tags\"]:\n",
    "        t = {\n",
    "            \"tag_name\": tag[\"name\"],\n",
    "            \"tag_display_name\": tag[\"display_name\"],\n",
    "            \"tag_vocabulary_id\": tag[\"vocabulary_id\"],\n",
    "        }\n",
    "        t.update(dataset)\n",
    "        t = pd.json_normalize(t, max_level=1)\n",
    "        tags_output.append(t)\n",
    "\n",
    "    for group in dataset[\"groups\"]:\n",
    "        g = {\n",
    "            \"group_name\": group[\"name\"],\n",
    "            \"group_display_name\": group[\"display_name\"],\n",
    "            \"group_title\": group[\"title\"],\n",
    "            \"group_display_name\": group[\"display_name\"],\n",
    "        }\n",
    "        g.update(dataset)\n",
    "        g = pd.json_normalize(g, max_level=1)\n",
    "        groups_output.append(g)\n",
    "\n",
    "    print(f\"Extracting resources for dataset: {dataset['title']}\")\n",
    "    resources = dataset.get_resources()\n",
    "    for resource in resources:\n",
    "        resource = extract_json_from_hdx_meta(resource)\n",
    "        resource.update(extract_json_from_hdx_meta(dataset))\n",
    "        resource = pd.json_normalize(resource, max_level=1)\n",
    "        resources_output.append(resource)\n",
    "\n",
    "print(\"Building output dataframe\")\n",
    "resources_output = pd.concat(resources_output)\n",
    "datasets_output = pd.DataFrame(datasets)\n",
    "tags_output = pd.concat(tags_output)\n",
    "groups_output = pd.concat(groups_output)\n",
    "\n",
    "# You might get some errors due to very long URLS, affecting a few rows\n",
    "print(\"Writing output to excel\")\n",
    "with pd.ExcelWriter(f\"./data/{country}_datasets_resources_tags_groups.xlsx\") as writer:\n",
    "    datasets_output.to_excel(writer, sheet_name=f\"Datasets\")\n",
    "    resources_output.to_excel(writer, sheet_name=f\"Dataset Resources\")\n",
    "    tags_output.to_excel(writer, sheet_name=f\"Dataset Tags\")\n",
    "    groups_output.to_excel(writer, sheet_name=f\"Dataset Groups\")\n",
    "\n",
    "print(\"Done\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Done\n"
     ]
    }
   ],
   "source": [
    "# Save to pkl files so we can easily start\n",
    "resources_output.to_pickle(\"./data/resources.pkl\")\n",
    "datasets_output.to_pickle(\"./data/datasets.pkl\")\n",
    "tags_output.to_pickle(\"./data/tags.pkl\")\n",
    "groups_output.to_pickle(\"./data/groups.pkl\")\n",
    "print(\"Done\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "We have 22089 datasets in HDX\n",
      "We have 148734 resources in HDX\n",
      "Index(['format', 'counts'], dtype='object') counts\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 2000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Top 100 tags ...\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>tag_name</th>\n",
       "      <th>counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>167</th>\n",
       "      <td>geodata</td>\n",
       "      <td>9132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>205</th>\n",
       "      <td>hxl</td>\n",
       "      <td>8416</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>210</th>\n",
       "      <td>indicators</td>\n",
       "      <td>6335</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>179</th>\n",
       "      <td>health</td>\n",
       "      <td>2432</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>376</th>\n",
       "      <td>transportation</td>\n",
       "      <td>2205</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>139</th>\n",
       "      <td>facilities and infrastructure</td>\n",
       "      <td>1976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>education</td>\n",
       "      <td>1959</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>baseline population</td>\n",
       "      <td>1834</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>116</th>\n",
       "      <td>economics</td>\n",
       "      <td>1700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>156</th>\n",
       "      <td>food security</td>\n",
       "      <td>1588</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>demographics</td>\n",
       "      <td>1239</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>96</th>\n",
       "      <td>development</td>\n",
       "      <td>1200</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>354</th>\n",
       "      <td>socioeconomics</td>\n",
       "      <td>1112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>402</th>\n",
       "      <td>violence and conflict</td>\n",
       "      <td>948</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>environment</td>\n",
       "      <td>786</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>221</th>\n",
       "      <td>integrated food security phase classification</td>\n",
       "      <td>727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>229</th>\n",
       "      <td>ipc</td>\n",
       "      <td>727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>186</th>\n",
       "      <td>historic ipc shapefiles</td>\n",
       "      <td>727</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>administrative divisions</td>\n",
       "      <td>689</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>181</th>\n",
       "      <td>health facilities</td>\n",
       "      <td>682</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61</th>\n",
       "      <td>children</td>\n",
       "      <td>680</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>140</th>\n",
       "      <td>facilities-infrastructure</td>\n",
       "      <td>667</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>airports</td>\n",
       "      <td>655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>common operational dataset - cod</td>\n",
       "      <td>646</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>agriculture</td>\n",
       "      <td>645</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>aviation</td>\n",
       "      <td>631</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>223</th>\n",
       "      <td>internally displaced persons - idp</td>\n",
       "      <td>559</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>304</th>\n",
       "      <td>populated places - settlements</td>\n",
       "      <td>558</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>105</th>\n",
       "      <td>displacement</td>\n",
       "      <td>539</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>333</th>\n",
       "      <td>roads</td>\n",
       "      <td>524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>282</th>\n",
       "      <td>nutrition</td>\n",
       "      <td>520</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>education facilities - schools</td>\n",
       "      <td>504</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>308</th>\n",
       "      <td>poverty</td>\n",
       "      <td>488</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129</th>\n",
       "      <td>energy</td>\n",
       "      <td>487</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>164</th>\n",
       "      <td>gender</td>\n",
       "      <td>485</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>405</th>\n",
       "      <td>water bodies - hydrography</td>\n",
       "      <td>452</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>373</th>\n",
       "      <td>trade</td>\n",
       "      <td>448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>152</th>\n",
       "      <td>floods - storm surges</td>\n",
       "      <td>448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>325</th>\n",
       "      <td>refugees</td>\n",
       "      <td>448</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>245</th>\n",
       "      <td>logistics</td>\n",
       "      <td>444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>264</th>\n",
       "      <td>mining</td>\n",
       "      <td>440</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>415</th>\n",
       "      <td>who is doing what and where - 3w - 4w - 5w</td>\n",
       "      <td>439</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>366</th>\n",
       "      <td>sustainable development goals-sdg</td>\n",
       "      <td>437</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>aid effectiveness</td>\n",
       "      <td>436</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>413</th>\n",
       "      <td>weather-climate</td>\n",
       "      <td>430</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>331</th>\n",
       "      <td>rivers</td>\n",
       "      <td>430</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>341</th>\n",
       "      <td>services</td>\n",
       "      <td>423</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>aid funding</td>\n",
       "      <td>414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>320</th>\n",
       "      <td>railways</td>\n",
       "      <td>404</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>307</th>\n",
       "      <td>ports</td>\n",
       "      <td>399</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>146</th>\n",
       "      <td>financial institutions</td>\n",
       "      <td>395</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>185</th>\n",
       "      <td>helicopter landing zone - hlz</td>\n",
       "      <td>387</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>299</th>\n",
       "      <td>points of interest - poi</td>\n",
       "      <td>386</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>buildings</td>\n",
       "      <td>384</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>178</th>\n",
       "      <td>hazards and risk</td>\n",
       "      <td>383</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>80</th>\n",
       "      <td>covid-19</td>\n",
       "      <td>382</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>310</th>\n",
       "      <td>prices</td>\n",
       "      <td>380</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>340</th>\n",
       "      <td>security incidents</td>\n",
       "      <td>374</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>316</th>\n",
       "      <td>protests</td>\n",
       "      <td>368</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>163</th>\n",
       "      <td>gazetteer</td>\n",
       "      <td>354</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>161</th>\n",
       "      <td>funding</td>\n",
       "      <td>334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>243</th>\n",
       "      <td>livelihoods</td>\n",
       "      <td>334</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>268</th>\n",
       "      <td>mortality</td>\n",
       "      <td>326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>86</th>\n",
       "      <td>cyclones - hurricanes - typhoons</td>\n",
       "      <td>312</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>275</th>\n",
       "      <td>needs assessment</td>\n",
       "      <td>300</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>disease</td>\n",
       "      <td>296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>current situation</td>\n",
       "      <td>296</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>256</th>\n",
       "      <td>medium term projection</td>\n",
       "      <td>280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>274</th>\n",
       "      <td>near term projection</td>\n",
       "      <td>280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>complex emergency</td>\n",
       "      <td>264</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>408</th>\n",
       "      <td>water sanitation and hygiene - wash</td>\n",
       "      <td>261</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>asylum seekers</td>\n",
       "      <td>250</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>distributions</td>\n",
       "      <td>249</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>101</th>\n",
       "      <td>disaster risk reduction - drr</td>\n",
       "      <td>244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>364</th>\n",
       "      <td>sustainable development</td>\n",
       "      <td>239</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>421</th>\n",
       "      <td>women</td>\n",
       "      <td>234</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>412</th>\n",
       "      <td>weather and climate</td>\n",
       "      <td>227</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>423</th>\n",
       "      <td>youth</td>\n",
       "      <td>225</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>casualties</td>\n",
       "      <td>225</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>262</th>\n",
       "      <td>millennium development goals - mdg</td>\n",
       "      <td>223</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>258</th>\n",
       "      <td>mental health</td>\n",
       "      <td>215</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>72</th>\n",
       "      <td>conflict-violence</td>\n",
       "      <td>214</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>246</th>\n",
       "      <td>malaria</td>\n",
       "      <td>212</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>elderly</td>\n",
       "      <td>209</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>362</th>\n",
       "      <td>survey</td>\n",
       "      <td>206</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>207</th>\n",
       "      <td>immunization</td>\n",
       "      <td>204</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>386</th>\n",
       "      <td>tuberculosis</td>\n",
       "      <td>202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>251</th>\n",
       "      <td>maternity</td>\n",
       "      <td>195</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>328</th>\n",
       "      <td>reproductive health and family planning</td>\n",
       "      <td>194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>112</th>\n",
       "      <td>earthquakes</td>\n",
       "      <td>192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>births</td>\n",
       "      <td>189</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>410</th>\n",
       "      <td>waterbodies</td>\n",
       "      <td>178</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>286</th>\n",
       "      <td>operational capacity</td>\n",
       "      <td>159</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>125</th>\n",
       "      <td>electricity</td>\n",
       "      <td>150</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>250</th>\n",
       "      <td>markets</td>\n",
       "      <td>149</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>416</th>\n",
       "      <td>who is doing what and where-3w-4w-5w</td>\n",
       "      <td>147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>access to water</td>\n",
       "      <td>141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>314</th>\n",
       "      <td>protection</td>\n",
       "      <td>141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>287</th>\n",
       "      <td>operational presence</td>\n",
       "      <td>136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>128</th>\n",
       "      <td>employment</td>\n",
       "      <td>131</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                          tag_name  counts\n",
       "167                                        geodata    9132\n",
       "205                                            hxl    8416\n",
       "210                                     indicators    6335\n",
       "179                                         health    2432\n",
       "376                                 transportation    2205\n",
       "139                  facilities and infrastructure    1976\n",
       "117                                      education    1959\n",
       "38                             baseline population    1834\n",
       "116                                      economics    1700\n",
       "156                                  food security    1588\n",
       "92                                    demographics    1239\n",
       "96                                     development    1200\n",
       "354                                 socioeconomics    1112\n",
       "402                          violence and conflict     948\n",
       "130                                    environment     786\n",
       "221  integrated food security phase classification     727\n",
       "229                                            ipc     727\n",
       "186                        historic ipc shapefiles     727\n",
       "11                        administrative divisions     689\n",
       "181                              health facilities     682\n",
       "61                                        children     680\n",
       "140                      facilities-infrastructure     667\n",
       "25                                        airports     655\n",
       "67                common operational dataset - cod     646\n",
       "19                                     agriculture     645\n",
       "35                                        aviation     631\n",
       "223             internally displaced persons - idp     559\n",
       "304                 populated places - settlements     558\n",
       "105                                   displacement     539\n",
       "333                                          roads     524\n",
       "282                                      nutrition     520\n",
       "119                 education facilities - schools     504\n",
       "308                                        poverty     488\n",
       "129                                         energy     487\n",
       "164                                         gender     485\n",
       "405                     water bodies - hydrography     452\n",
       "373                                          trade     448\n",
       "152                          floods - storm surges     448\n",
       "325                                       refugees     448\n",
       "245                                      logistics     444\n",
       "264                                         mining     440\n",
       "415     who is doing what and where - 3w - 4w - 5w     439\n",
       "366              sustainable development goals-sdg     437\n",
       "20                               aid effectiveness     436\n",
       "413                                weather-climate     430\n",
       "331                                         rivers     430\n",
       "341                                       services     423\n",
       "21                                     aid funding     414\n",
       "320                                       railways     404\n",
       "307                                          ports     399\n",
       "146                         financial institutions     395\n",
       "185                  helicopter landing zone - hlz     387\n",
       "299                       points of interest - poi     386\n",
       "44                                       buildings     384\n",
       "178                               hazards and risk     383\n",
       "80                                        covid-19     382\n",
       "310                                         prices     380\n",
       "340                             security incidents     374\n",
       "316                                       protests     368\n",
       "163                                      gazetteer     354\n",
       "161                                        funding     334\n",
       "243                                    livelihoods     334\n",
       "268                                      mortality     326\n",
       "86                cyclones - hurricanes - typhoons     312\n",
       "275                               needs assessment     300\n",
       "103                                        disease     296\n",
       "85                               current situation     296\n",
       "256                        medium term projection      280\n",
       "274                          near term projection      280\n",
       "69                               complex emergency     264\n",
       "408            water sanitation and hygiene - wash     261\n",
       "32                                  asylum seekers     250\n",
       "106                                  distributions     249\n",
       "101                  disaster risk reduction - drr     244\n",
       "364                        sustainable development     239\n",
       "421                                          women     234\n",
       "412                            weather and climate     227\n",
       "423                                          youth     225\n",
       "52                                      casualties     225\n",
       "262             millennium development goals - mdg     223\n",
       "258                                  mental health     215\n",
       "72                               conflict-violence     214\n",
       "246                                        malaria     212\n",
       "123                                        elderly     209\n",
       "362                                         survey     206\n",
       "207                                   immunization     204\n",
       "386                                   tuberculosis     202\n",
       "251                                      maternity     195\n",
       "328        reproductive health and family planning     194\n",
       "112                                    earthquakes     192\n",
       "40                                          births     189\n",
       "410                                    waterbodies     178\n",
       "286                           operational capacity     159\n",
       "125                                    electricity     150\n",
       "250                                        markets     149\n",
       "416           who is doing what and where-3w-4w-5w     147\n",
       "4                                  access to water     141\n",
       "314                                     protection     141\n",
       "287                           operational presence     136\n",
       "128                                     employment     131"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Top 100 groups ...\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>group_display_name</th>\n",
       "      <th>counts</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>238</th>\n",
       "      <td>United States</td>\n",
       "      <td>654</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>248</th>\n",
       "      <td>World</td>\n",
       "      <td>515</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>119</th>\n",
       "      <td>Kenya</td>\n",
       "      <td>454</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50</th>\n",
       "      <td>Colombia</td>\n",
       "      <td>414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74</th>\n",
       "      <td>Ethiopia</td>\n",
       "      <td>375</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>62</th>\n",
       "      <td>Democratic Republic of the Congo</td>\n",
       "      <td>371</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>210</th>\n",
       "      <td>South Sudan</td>\n",
       "      <td>365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>207</th>\n",
       "      <td>Somalia</td>\n",
       "      <td>362</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>233</th>\n",
       "      <td>Uganda</td>\n",
       "      <td>347</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Afghanistan</td>\n",
       "      <td>337</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>162</th>\n",
       "      <td>Nigeria</td>\n",
       "      <td>326</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>214</th>\n",
       "      <td>Sudan</td>\n",
       "      <td>325</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>Bangladesh</td>\n",
       "      <td>309</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>151</th>\n",
       "      <td>Mozambique</td>\n",
       "      <td>309</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>137</th>\n",
       "      <td>Mali</td>\n",
       "      <td>299</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>175</th>\n",
       "      <td>Philippines</td>\n",
       "      <td>297</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>43</th>\n",
       "      <td>Chad</td>\n",
       "      <td>291</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>161</th>\n",
       "      <td>Niger</td>\n",
       "      <td>289</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>Burkina Faso</td>\n",
       "      <td>287</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>219</th>\n",
       "      <td>Syrian Arab Republic</td>\n",
       "      <td>281</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>155</th>\n",
       "      <td>Nepal</td>\n",
       "      <td>274</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>107</th>\n",
       "      <td>Indonesia</td>\n",
       "      <td>265</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>100</th>\n",
       "      <td>Haiti</td>\n",
       "      <td>245</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>42</th>\n",
       "      <td>Central African Republic</td>\n",
       "      <td>244</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79</th>\n",
       "      <td>France</td>\n",
       "      <td>240</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>141</th>\n",
       "      <td>Mauritania</td>\n",
       "      <td>238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>134</th>\n",
       "      <td>Malawi</td>\n",
       "      <td>236</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>39</th>\n",
       "      <td>Cameroon</td>\n",
       "      <td>235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>152</th>\n",
       "      <td>Myanmar</td>\n",
       "      <td>235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>Burundi</td>\n",
       "      <td>233</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>251</th>\n",
       "      <td>Zimbabwe</td>\n",
       "      <td>230</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>133</th>\n",
       "      <td>Madagascar</td>\n",
       "      <td>222</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>Ecuador</td>\n",
       "      <td>211</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>198</th>\n",
       "      <td>Senegal</td>\n",
       "      <td>208</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>109</th>\n",
       "      <td>Iraq</td>\n",
       "      <td>206</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>169</th>\n",
       "      <td>Pakistan</td>\n",
       "      <td>204</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>95</th>\n",
       "      <td>Guatemala</td>\n",
       "      <td>204</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>97</th>\n",
       "      <td>Guinea</td>\n",
       "      <td>203</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>185</th>\n",
       "      <td>Rwanda</td>\n",
       "      <td>199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>249</th>\n",
       "      <td>Yemen</td>\n",
       "      <td>197</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>128</th>\n",
       "      <td>Liberia</td>\n",
       "      <td>194</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>250</th>\n",
       "      <td>Zambia</td>\n",
       "      <td>192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>103</th>\n",
       "      <td>Honduras</td>\n",
       "      <td>192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>237</th>\n",
       "      <td>United Republic of Tanzania</td>\n",
       "      <td>188</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>201</th>\n",
       "      <td>Sierra Leone</td>\n",
       "      <td>187</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40</th>\n",
       "      <td>Canada</td>\n",
       "      <td>179</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>64</th>\n",
       "      <td>Djibouti</td>\n",
       "      <td>168</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>Brazil</td>\n",
       "      <td>167</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Angola</td>\n",
       "      <td>165</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>234</th>\n",
       "      <td>Ukraine</td>\n",
       "      <td>164</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>213</th>\n",
       "      <td>State of Palestine</td>\n",
       "      <td>158</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>245</th>\n",
       "      <td>Viet Nam</td>\n",
       "      <td>152</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>88</th>\n",
       "      <td>Ghana</td>\n",
       "      <td>148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>127</th>\n",
       "      <td>Lesotho</td>\n",
       "      <td>147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>60</th>\n",
       "      <td>Côte d'Ivoire</td>\n",
       "      <td>147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>Benin</td>\n",
       "      <td>147</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>85</th>\n",
       "      <td>Gambia</td>\n",
       "      <td>144</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>China</td>\n",
       "      <td>143</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>208</th>\n",
       "      <td>South Africa</td>\n",
       "      <td>142</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>129</th>\n",
       "      <td>Libya</td>\n",
       "      <td>141</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>126</th>\n",
       "      <td>Lebanon</td>\n",
       "      <td>140</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>115</th>\n",
       "      <td>Japan</td>\n",
       "      <td>138</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>144</th>\n",
       "      <td>Mexico</td>\n",
       "      <td>137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>174</th>\n",
       "      <td>Peru</td>\n",
       "      <td>136</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>52</th>\n",
       "      <td>Congo</td>\n",
       "      <td>135</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>106</th>\n",
       "      <td>India</td>\n",
       "      <td>132</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Algeria</td>\n",
       "      <td>131</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>44</th>\n",
       "      <td>Chile</td>\n",
       "      <td>129</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>244</th>\n",
       "      <td>Venezuela (Bolivarian Republic of)</td>\n",
       "      <td>127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>153</th>\n",
       "      <td>Namibia</td>\n",
       "      <td>127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>38</th>\n",
       "      <td>Cambodia</td>\n",
       "      <td>127</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>224</th>\n",
       "      <td>Togo</td>\n",
       "      <td>126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>113</th>\n",
       "      <td>Italy</td>\n",
       "      <td>126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>Eritrea</td>\n",
       "      <td>126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>232</th>\n",
       "      <td>Türkiye</td>\n",
       "      <td>126</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>98</th>\n",
       "      <td>Guinea-Bissau</td>\n",
       "      <td>120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>69</th>\n",
       "      <td>El Salvador</td>\n",
       "      <td>120</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>177</th>\n",
       "      <td>Poland</td>\n",
       "      <td>119</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>123</th>\n",
       "      <td>Kyrgyzstan</td>\n",
       "      <td>118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>66</th>\n",
       "      <td>Dominican Republic</td>\n",
       "      <td>118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>236</th>\n",
       "      <td>United Kingdom</td>\n",
       "      <td>118</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>Bolivia (Plurinational State of)</td>\n",
       "      <td>115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>Egypt</td>\n",
       "      <td>115</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>Eswatini</td>\n",
       "      <td>113</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>Botswana</td>\n",
       "      <td>112</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>117</th>\n",
       "      <td>Jordan</td>\n",
       "      <td>111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>84</th>\n",
       "      <td>Gabon</td>\n",
       "      <td>111</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>228</th>\n",
       "      <td>Tunisia</td>\n",
       "      <td>109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>51</th>\n",
       "      <td>Comoros</td>\n",
       "      <td>109</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>150</th>\n",
       "      <td>Morocco</td>\n",
       "      <td>108</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>157</th>\n",
       "      <td>Netherlands</td>\n",
       "      <td>106</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>Cabo Verde</td>\n",
       "      <td>104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>160</th>\n",
       "      <td>Nicaragua</td>\n",
       "      <td>104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>Equatorial Guinea</td>\n",
       "      <td>104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>124</th>\n",
       "      <td>Lao People's Democratic Republic</td>\n",
       "      <td>103</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>196</th>\n",
       "      <td>Sao Tome and Principe</td>\n",
       "      <td>102</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>221</th>\n",
       "      <td>Tajikistan</td>\n",
       "      <td>101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>222</th>\n",
       "      <td>Thailand</td>\n",
       "      <td>101</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>142</th>\n",
       "      <td>Mauritius</td>\n",
       "      <td>99</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>172</th>\n",
       "      <td>Papua New Guinea</td>\n",
       "      <td>98</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     group_display_name  counts\n",
       "238                       United States     654\n",
       "248                               World     515\n",
       "119                               Kenya     454\n",
       "50                             Colombia     414\n",
       "74                             Ethiopia     375\n",
       "62     Democratic Republic of the Congo     371\n",
       "210                         South Sudan     365\n",
       "207                             Somalia     362\n",
       "233                              Uganda     347\n",
       "0                           Afghanistan     337\n",
       "162                             Nigeria     326\n",
       "214                               Sudan     325\n",
       "17                           Bangladesh     309\n",
       "151                          Mozambique     309\n",
       "137                                Mali     299\n",
       "175                         Philippines     297\n",
       "43                                 Chad     291\n",
       "161                               Niger     289\n",
       "35                         Burkina Faso     287\n",
       "219                Syrian Arab Republic     281\n",
       "155                               Nepal     274\n",
       "107                           Indonesia     265\n",
       "100                               Haiti     245\n",
       "42             Central African Republic     244\n",
       "79                               France     240\n",
       "141                          Mauritania     238\n",
       "134                              Malawi     236\n",
       "39                             Cameroon     235\n",
       "152                             Myanmar     235\n",
       "36                              Burundi     233\n",
       "251                            Zimbabwe     230\n",
       "133                          Madagascar     222\n",
       "67                              Ecuador     211\n",
       "198                             Senegal     208\n",
       "109                                Iraq     206\n",
       "169                            Pakistan     204\n",
       "95                            Guatemala     204\n",
       "97                               Guinea     203\n",
       "185                              Rwanda     199\n",
       "249                               Yemen     197\n",
       "128                             Liberia     194\n",
       "250                              Zambia     192\n",
       "103                            Honduras     192\n",
       "237         United Republic of Tanzania     188\n",
       "201                        Sierra Leone     187\n",
       "40                               Canada     179\n",
       "64                             Djibouti     168\n",
       "30                               Brazil     167\n",
       "5                                Angola     165\n",
       "234                             Ukraine     164\n",
       "213                  State of Palestine     158\n",
       "245                            Viet Nam     152\n",
       "88                                Ghana     148\n",
       "127                             Lesotho     147\n",
       "60                        Côte d'Ivoire     147\n",
       "22                                Benin     147\n",
       "85                               Gambia     144\n",
       "45                                China     143\n",
       "208                        South Africa     142\n",
       "129                               Libya     141\n",
       "126                             Lebanon     140\n",
       "115                               Japan     138\n",
       "144                              Mexico     137\n",
       "174                                Peru     136\n",
       "52                                Congo     135\n",
       "106                               India     132\n",
       "2                               Algeria     131\n",
       "44                                Chile     129\n",
       "244  Venezuela (Bolivarian Republic of)     127\n",
       "153                             Namibia     127\n",
       "38                             Cambodia     127\n",
       "224                                Togo     126\n",
       "113                               Italy     126\n",
       "71                              Eritrea     126\n",
       "232                             Türkiye     126\n",
       "98                        Guinea-Bissau     120\n",
       "69                          El Salvador     120\n",
       "177                              Poland     119\n",
       "123                          Kyrgyzstan     118\n",
       "66                   Dominican Republic     118\n",
       "236                      United Kingdom     118\n",
       "25     Bolivia (Plurinational State of)     115\n",
       "68                                Egypt     115\n",
       "73                             Eswatini     113\n",
       "28                             Botswana     112\n",
       "117                              Jordan     111\n",
       "84                                Gabon     111\n",
       "228                             Tunisia     109\n",
       "51                              Comoros     109\n",
       "150                             Morocco     108\n",
       "157                         Netherlands     106\n",
       "37                           Cabo Verde     104\n",
       "160                           Nicaragua     104\n",
       "70                    Equatorial Guinea     104\n",
       "124    Lao People's Democratic Republic     103\n",
       "196               Sao Tome and Principe     102\n",
       "221                          Tajikistan     101\n",
       "222                            Thailand     101\n",
       "142                           Mauritius      99\n",
       "172                    Papua New Guinea      98"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#resources_output = pd.read_pickle('./data/resources.pkl')\n",
    "#datasets_output = pd.read_pickle('./data/datasets.pkl')\n",
    "#tags_output = pd.read_pickle('./data/tags.pkl')\n",
    "#groups_output = pd.read_pickle('./data/groups.pkl')\n",
    "\n",
    "\n",
    "def plot_sorted_bar_chart(\n",
    "    df, x, y, title, xlabel, ylabel, figsize=(10, 10), rotation=0\n",
    "):\n",
    "    \"\"\"\n",
    "    Utility function for simple bar plots\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    df : pandas.DataFrame\n",
    "        The dataframe to be plotted.\n",
    "    x : str\n",
    "        The column name of the x-axis.\n",
    "    y : str\n",
    "        The column name of the y-axis.\n",
    "    title : str\n",
    "        The title of the plot.\n",
    "    xlabel : str\n",
    "        The label of the x-axis.\n",
    "    ylabel : str\n",
    "        The label of the y-axis.\n",
    "    figsize : tuple\n",
    "        The size of the figure.\n",
    "    rotation : int\n",
    "        The rotation of the x-axis labels.\n",
    "\n",
    "    \"\"\"\n",
    "    totals = df.groupby([x]).size().reset_index(name=y)\n",
    "    print(totals.columns, y)\n",
    "    totals = totals.sort_values(by=y, ascending=False)\n",
    "    totals.plot.bar(x=x, y=y, figsize=figsize, rot=rotation, fontsize=16, legend=False)\n",
    "    plt.rcParams.update({'font.size': 22}) \n",
    "    plt.title(title)\n",
    "    plt.xlabel(xlabel)\n",
    "    plt.ylabel(ylabel)\n",
    "    plt.show()\n",
    "\n",
    "\n",
    "def display_totals(df, x, y, n):\n",
    "    \"\"\"\n",
    "    This function takes a dataframe, a column name, a column name, and an integer.\n",
    "    It returns a dataframe with the top n rows of the dataframe sorted by the column name.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    df : dataframe\n",
    "        The dataframe to be sorted.\n",
    "    x : string\n",
    "        The column name to be used to group data\n",
    "    y : string\n",
    "        Name of the count column\n",
    "\n",
    "    \"\"\"\n",
    "    totals = df.groupby([x]).size().reset_index(name=y)\n",
    "    totals = totals.sort_values(by=y, ascending=False)\n",
    "    display(totals[0:n])\n",
    "\n",
    "\n",
    "print(\"We have {} datasets in HDX\".format(len(datasets_output)))\n",
    "print(\"We have {} resources in HDX\".format(len(resources_output)))\n",
    "\n",
    "plot_sorted_bar_chart(\n",
    "    resources_output,\n",
    "    \"format\",\n",
    "    \"counts\",\n",
    "    \"Resource File Types on the HDX Platform\",\n",
    "    \"File type\",\n",
    "    \"Files\",\n",
    "    rotation=90,\n",
    "    figsize=(20, 10),\n",
    ")\n",
    "\n",
    "print(\"\\nTop 100 tags ...\")\n",
    "display_totals(tags_output, \"tag_name\", \"counts\", 100)\n",
    "\n",
    "print(\"\\nTop 100 groups ...\")\n",
    "display_totals(groups_output, \"group_display_name\", \"counts\", 100)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Download data files from HDX\n",
    "\n",
    "So we can interrogate files to get column names, field level HDX tags and data sample, we need to download them. This download is imperfect at this time, known issues are:\n",
    "\n",
    "1. Zip files - we leave unpacked for now will come back to this\n",
    "2. Timeouts - large files seem to timeout, couldn't see a way to adjust this in python module\n",
    "\n",
    "Due to the above the following following panel will need to be run a few times. Since the aim of this analysis is to get a reasonable set of tabular data for explorations, this should suffice for now. But some work in the future to harden this process. I suspect the best way forward for full control would be to hit the [REST API](https://docs.ckan.org/en/2.9/api/) directly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "def check_hdx_header(first_row):\n",
    "    \"\"\"\n",
    "    This function checks if the first row of a csv file likely an HDX header.\n",
    "    \"\"\"\n",
    "    matches = [\"#meta\", \"#country\", \"#data\", \"#loc\", \"#geo\"]\n",
    "    if any(x in first_row for x in matches):\n",
    "        return True\n",
    "    else:\n",
    "        return False\n",
    "\n",
    "\n",
    "def set_meta_data_fields(data, file, dataset, resource, sheet, type):\n",
    "\n",
    "    \"\"\"\n",
    "    This function create a data frame with meta data about the data, as well as a snippet of its\n",
    "    first nrows.\n",
    "\n",
    "    Parameters:\n",
    "        data: a dataframe\n",
    "        file: the name of the data file\n",
    "        dataset: the dataset JSON object from HDX\n",
    "        resource: the resource JSON object from HDX\n",
    "        sheet: the sheet name if the data was a tab in a sheet\n",
    "        type: the type of file, CSV, XLSX, etc.\n",
    "    Returns:\n",
    "        dict: a dictionary with metadata about the dataframe\n",
    "    \"\"\"\n",
    "\n",
    "    nrows = 10\n",
    "\n",
    "    # Data preview to only include values\n",
    "    data = data.dropna(axis=1, how=\"all\")\n",
    "\n",
    "    cols = str(list(data.columns))\n",
    "    if data.shape[0] > 0:\n",
    "        first_row = str(list(data.iloc[0]))\n",
    "        has_hxl_header = check_hdx_header(first_row)\n",
    "        num_rows = int(data.shape[0])\n",
    "        num_cols = int(data.shape[1])\n",
    "        first_nrows = data.head(nrows)\n",
    "    else:\n",
    "        first_row = \"No data\"\n",
    "        has_hxl_header = \"No data\"\n",
    "        num_rows = 0\n",
    "        num_cols = 0\n",
    "        first_nrows = None\n",
    "\n",
    "    dict = {}\n",
    "\n",
    "    dict[\"resource_id\"] = resource[\"id\"]\n",
    "    dict[\"resource_name\"] = resource[\"name\"]\n",
    "    dict[\"resource_format\"] = resource[\"format\"]\n",
    "    dict[\"dataset_id\"] = dataset[\"id\"]\n",
    "    dict[\"dataset_name\"] = dataset[\"name\"]\n",
    "    dict[\"dataset_org_title\"] = dataset[\"organization\"][\"title\"]\n",
    "    dict[\"dataset_last_modified\"] = dataset[\"last_modified\"]\n",
    "    dict[\"dataset_tags\"] = dataset[\"tags\"]\n",
    "    dict[\"dataset_groups\"] = dataset[\"groups\"]\n",
    "    dict[\"dataset_total_res_downloads\"] = dataset[\"total_res_downloads\"]\n",
    "    dict[\"dataset_pageviews_last_14_days\"] = dataset[\"pageviews_last_14_days\"]\n",
    "    dict[\"file\"] = file\n",
    "    dict[\"type\"] = type\n",
    "    dict[\"dataset\"] = dataset\n",
    "    dict[\"sheet\"] = sheet\n",
    "    dict[\"resource\"] = resource\n",
    "    dict[\"num_rows\"] = num_rows\n",
    "    dict[\"num_cols\"] = num_cols\n",
    "    dict[\"columns\"] = cols\n",
    "    dict[\"first_row\"] = first_row\n",
    "    dict[\"has_hxl_header\"] = has_hxl_header\n",
    "    dict[\"first_nrows\"] = first_nrows\n",
    "    return dict\n",
    "\n",
    "\n",
    "def extract_data_details(f, dataset, resource, nrows, data_details):\n",
    "    \"\"\"\n",
    "\n",
    "    Reads saved CVS and XLSX HDX files and extracts headers, HDX tags and sample data.\n",
    "    For XLSX files, it extracts data from all sheets.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    f : str\n",
    "        The file name\n",
    "    dataset : str\n",
    "        The dataset name\n",
    "    resource : str\n",
    "        The resource name\n",
    "    nrows : int\n",
    "        The number of rows to read\n",
    "    data_details : list\n",
    "        The list of data details\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    data_details : list\n",
    "        The list of data details\n",
    "\n",
    "    \"\"\"\n",
    "    if f.endswith(\".xlsx\") or f.endswith(\".xls\"):\n",
    "        print(f\"Loading xslx file {f} ...\")\n",
    "        try:\n",
    "            sheet_to_df_map = pd.read_excel(f, sheet_name=None)\n",
    "        except Exception:\n",
    "            print(\"An exception occurred trying to read the file {f}\")\n",
    "            return data_details\n",
    "        for sheet in sheet_to_df_map:\n",
    "            data = sheet_to_df_map[sheet]\n",
    "            data_details.append(\n",
    "                set_meta_data_fields(data, f, dataset, resource, sheet, \"xlsx\")\n",
    "            )\n",
    "    elif f.endswith(\".csv\"):\n",
    "        print(f\"Loading csv file {f}\")\n",
    "        # Detect encoding\n",
    "        with open(f, \"rb\") as rawdata:\n",
    "            r = chardet.detect(rawdata.read(100000))\n",
    "        try:\n",
    "            data = pd.read_csv(f, encoding=r[\"encoding\"], encoding_errors=\"ignore\")\n",
    "        except Exception:\n",
    "            print(\"An exception occurred trying to read the file {f}\")\n",
    "            return data_details\n",
    "        data_details.append(set_meta_data_fields(data, f, dataset, resource, \"\", \"csv\"))\n",
    "    else:\n",
    "        type = f.split(\".\")[-1]\n",
    "        print(f\"Type {type} for {f}\")\n",
    "        data = pd.DataFrame()\n",
    "        data_details.append(set_meta_data_fields(data, f, dataset, resource, \"\", type))\n",
    "\n",
    "    return data_details\n",
    "\n",
    "\n",
    "# Loop through downloaded folders\n",
    "def extract_all_data_details(startpath, data_details):\n",
    "    \"\"\"\n",
    "    Extracts all data details from all datasets in the given directory.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    startpath : str\n",
    "        The path to the directory containing all datasets.\n",
    "    data_details : list\n",
    "        Results\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    data_details : pandas.DataFrame\n",
    "        Results, to which new meta data was appended\n",
    "    \"\"\"\n",
    "    for d in os.listdir(startpath):\n",
    "        if \".DS_\" not in d:\n",
    "            d = f\"{startpath}/{d}\"\n",
    "            with open(f\"{d}/dataset.json\") as f:\n",
    "                dataset = json.load(f)\n",
    "            for r in os.listdir(d):\n",
    "                if \"dataset.json\" not in r and \".DS_\" not in r:\n",
    "                    with open(f\"{d}/{r}/resource.json\") as f:\n",
    "                        resource = json.load(f)\n",
    "                    for f in os.listdir(f\"{d}/{r}\"):\n",
    "                        file = str(f\"{d}/{r}/{f}\")\n",
    "                        if \".json\" not in file and \".DS_\" not in file:\n",
    "                            data_details = extract_data_details(\n",
    "                                file, dataset, resource, 5, data_details\n",
    "                            )\n",
    "    data_details = pd.DataFrame(data_details)\n",
    "    return data_details\n",
    "\n",
    "\n",
    "def is_supported_filetype(format):\n",
    "    \"\"\"\n",
    "    Checks if the file format is currently supported for extracting meta data.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    format : str\n",
    "        The file format to check.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    bool\n",
    "        True if the file format is supported, False otherwise.\n",
    "    \"\"\"\n",
    "    matches = [\"CSV\", \"XLSX\", \"XLS\", \"ZIP\", \"TSV\"]\n",
    "    if any(x in format for x in matches):\n",
    "        return True\n",
    "    else:\n",
    "        return False\n",
    "\n",
    "\n",
    "def download_data(datasets, output_folder):\n",
    "    \"\"\"\n",
    "    Downloads data from HDX. Will save dataset and resource meta data for each file\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    datasets : pandas.DataFrame\n",
    "        A dataframe containing the datasets to download.\n",
    "    output_folder : str\n",
    "        The folder to download the data to.\n",
    "\n",
    "    \"\"\"\n",
    "    if not os.path.exists(output_folder):\n",
    "        os.mkdir(output_folder)\n",
    "\n",
    "    for index, row in datasets.iterrows():\n",
    "        dataset = Dataset.read_from_hdx(row[\"id\"])\n",
    "        resources = dataset.get_resources()\n",
    "        for resource in resources:\n",
    "            dir = f\"./{output_folder}/{row['name']}_{row['id']}\"\n",
    "            print(\n",
    "                f\"Downloading {row['name']} - {resource['name']} - {resource['format']}\"\n",
    "            )\n",
    "            resource[\"dataset_name\"] = row[\"name\"]\n",
    "            if not os.path.exists(dir):\n",
    "                dump_hdx_meta_file(dataset, dir, \"dataset.json\")\n",
    "            # We need a try/catch because sometimes the HDX API times out\n",
    "            try:\n",
    "                dir = f'{dir}/{get_safe_name(resource[\"name\"])}_{get_safe_name(resource[\"id\"])}'\n",
    "                if not os.path.exists(dir):\n",
    "                    dump_hdx_meta_file(resource, dir, \"resource.json\")\n",
    "                    if is_supported_filetype(resource[\"format\"]):\n",
    "                        url, path = resource.download(dir)\n",
    "                        # Unzip zipped files. TODO actually likely won't work as HDX zip files don't end with\n",
    "                        # .zip, but rather have zip in name, for example file.zip.shp will actually be a zip file\n",
    "                        # with multiple files. Will come back to thia.\n",
    "                        if path.endswith(\".zip\"):\n",
    "                            print(f\"Unzipping {path}\")\n",
    "                            with zipfile.ZipFile(path, \"r\") as zip_ref:\n",
    "                                zip_ref.extractall(dir)\n",
    "                    else:\n",
    "                        # TODO Add support for skipped file types\n",
    "                        print(\n",
    "                            f\"*** Skipping file as it is not a supported filetype *** {resource['name']}\"\n",
    "                        )\n",
    "                else:\n",
    "                    print(f\"Skipping {dir} as it already exists\")\n",
    "            except Exception as e:\n",
    "                traceback.print_exc()\n",
    "                print(\n",
    "                    f\"An exception occurred trying to analyze the dataset {row['name']}\"\n",
    "                )\n",
    "                sys.exit()\n",
    "\n",
    "    print(\"Done\")\n",
    "\n",
    "\n",
    "def extract_json_from_hdx_meta(hdx_object):\n",
    "    \"\"\"\n",
    "    This function takes a HDX object as input and returns a JSON object.\n",
    "\n",
    "    Parameters:\n",
    "    hdx_object (dict): A HDX object\n",
    "\n",
    "    Returns:\n",
    "    json_op (str): A JSON object\n",
    "    \"\"\"\n",
    "    json_op = {}\n",
    "    for key in hdx_object:\n",
    "        json_op[key] = hdx_object[key]\n",
    "    json_op = json.dumps(json_op, indent=4)\n",
    "    return json_op\n",
    "\n",
    "\n",
    "def dump_hdx_meta_file(hdx_object, dir, filename):\n",
    "    \"\"\"\n",
    "    This function takes in a HDX object and dumps the metadata in a json file.\n",
    "\n",
    "    Parameters:\n",
    "\n",
    "    hdx_object (object): HDX object\n",
    "    dir (str): Directory where the json file is to be dumped\n",
    "    filename (str): Name of the json file\n",
    "\n",
    "    Returns:\n",
    "    None\n",
    "    \"\"\"\n",
    "    if not os.path.exists(dir):\n",
    "        os.mkdir(dir)\n",
    "    json_op = extract_json_from_hdx_meta(hdx_object)\n",
    "    with open(f\"{dir}/{filename}\", \"w\") as f:\n",
    "        f.write(json_op)\n",
    "\n",
    "\n",
    "def get_safe_name(name):\n",
    "    \"\"\"\n",
    "    This function takes a string and returns a version of it that is safe to use as a filename.\n",
    "\n",
    "    Parameters:\n",
    "        name (str): The string to be converted to a safe filename.\n",
    "\n",
    "    Returns:\n",
    "        str: The safe filename.\n",
    "    \"\"\"\n",
    "    return re.sub(\"[^0-9a-zA-Z]+\", \"_\", name)\n",
    "\n",
    "\n",
    "def list_files(startpath):\n",
    "    \"\"\"\n",
    "    lists all the files in a path, with some indenting.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    startpath : str\n",
    "        The path to the directory to be listed.\n",
    "\n",
    "    \"\"\"\n",
    "    for root, dirs, files in os.walk(startpath):\n",
    "        level = root.replace(startpath, \"\").count(os.sep)\n",
    "        indent = \" \" * 4 * (level)\n",
    "        print(\"{}{}/\".format(indent, os.path.basename(root)))\n",
    "        subindent = \" \" * 4 * (level + 1)\n",
    "        for f in files:\n",
    "            print(\"{}{}\".format(subindent, f))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##  Download Kenya Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets = pd.read_pickle(\"./data/datasets.pkl\")\n",
    "\n",
    "# Subset to {country} as set above\n",
    "datasets_subset = datasets.loc[datasets[\"groups\"].astype(str).str.contains(country)]\n",
    "\n",
    "print(f\"Found {len(datasets_subset)} datasets for {country} in HDX\")\n",
    "download_data(datasets_subset, f\"./data/{country}\", agent_count)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Build Metadata summary for Kenya with details extracted from data files\n",
    "\n",
    "We will scan through the downloaded files and extract data and meta data. This extract will process csv data files, plus all sheets in Excel files. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "resources_data_details = extract_all_data_details(f\"./data/{country}\", [])\n",
    "resources_data_details.to_pickle(f\"./data/{country}_data_details.pkl\")\n",
    "print(\"Done\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check how many of our Kenyan files have HXL tags"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Index(['has_hxl_header', 'counts'], dtype='object') counts\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAABocAAAO+CAYAAADGz8yxAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/NK7nSAAAACXBIWXMAAA9hAAAPYQGoP6dpAACsIklEQVR4nOzdeZQU1f0/7nfPAMMOIogLqyigoEZExQXFBSOiuG9EAaNRg0ZNTIxojCaiaPwaNW5Ro6DGfSFGccWIC7iDIFFABERENtkZZO3fH/ymP9PMjjMOpJ7nnD50Vd1761Z3TzWnXl33ptLpdDoAAAAAAABIhJzq7gAAAAAAAAA/HuEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAACwSebOnRtXX3117LfffrH11ltHjRo1IpVKRSqVih49emTKDRs2LLO+TZs2JbY3atSoTLlUKlX1B0BERNZrPmrUqEpvf8CAAZn2BwwYUOntQ9KU95wKAFCaGtXdAQAAYMvzzjvvxHHHHRffffdddXcFAACACnLnEACQCIV/ZVvRX8f/61//iry8vEzdGjVqxMMPP1x1neUH69GjR9b7vfEjLy8vGjduHDvvvHP06NEjzj///Lj//vtj+vTp1d31LDNmzIhrrrkm89hcLF26NE488cSsYKh+/frRtm3baNeuXbRr1y522GGHauwhVeHWW2/NfBY/+eST6u7OJrnmmmuyzgUzZsyocBuFzy9l3bXRpk2bUs9FtWvXjiZNmkSHDh3i8MMPj1/96lfxyCOPxLffflvhfvXs2TPrHPff//63QvXfe++9yM3NzbRx8sknV7gPxSl8vJty59jGdxQOGzasUvoFAJB07hwCACjFI488EgMGDIi1a9dGREStWrXiscceixNOOKGae8YPsXr16li9enUsWbIkpk6dGm+++Wbcc889ERGxzz77xEUXXRSnnXZa5ObmVms/Z8yYEX/6058yy5tLQPTwww/HvHnzIiKiTp068fjjj8cxxxxjKLj/cbfeemt89dVXEbEh9PjJT35SvR36H7Bq1apYtWpVLFq0KKZMmRKvv/563HHHHZGTkxOHHHJIXHLJJXH00UeXq6377rsvdtttt1i+fHmsXr06fv7zn8eYMWPKdR77/vvv46yzzor169dHRETTpk3jzjvv/EHHBgDA5s2dQwAAJbj33nujX79+mWCoTp068e9//1swtIWpUaNG5m6Wgkfr1q2jYcOGxYYZH3zwQZxxxhmxzz77VPiX90nxn//8J/P8zDPPjD59+pQaDA0YMCDS6XSk0+lNulODqlXw3qTT6ay5oqhctWvXLnIuatmyZdSvX79I2fXr18frr78exxxzTBx22GExa9asMttv06ZN3HjjjZnlDz74IP7f//t/5erbH//4x5g0aVJm+fbbb49tttmmXHUBANgyCYcAAIrx17/+Nc4777zMr6gbNGgQL7/8cvz0pz+t5p5RUTvssENMnTo16zFjxoxYsmRJrF27Nj799NO47777olevXlkBx9ixY2PfffeNMWPGVGPvN0/Tpk3LPN9jjz2qsSew5dh3332LnItmzpwZy5Yti9WrV8dHH30Ud9xxR+y///5Z9f7zn//EXnvtFV988UWZ+/jlL38ZBx98cGb5mmuuyQp9ivP+++/HX//618zycccdF6eddloFjw4AgC2NcAgAYCN//vOf49JLL80sN2nSJF5//fU46KCDqrFXVIWcnJzo3LlznHPOOfHiiy/Gf//73+jWrVtm+4oVK+K4447b7OYiqm5Lly7NPK9bt2419gT+N9SsWTP22muvuOCCC2L06NHx7rvvRvv27TPb582bF717946FCxeW2k4qlYr7778/83f5/fffx89//vPMDx02tmrVqjjrrLNi3bp1EbHh++7uu++upKMCAGBzZs4hAIBCLrvssrjpppsyy82bN4/XXnstdtttt3K3sX79+vjggw9i8uTJMXfu3KhZs2Zst9120b1799hhhx0qpZ+ff/55jB07NmbNmhV16tSJtm3bxiGHHFLs8ETVZe7cufHpp5/G1KlTY/HixZGTkxNbb711dOzYMfbZZ5+oWbNmdXexiF122SXeeuut6NevXzz++OMRETF//vzo169fvP322+VqY/ny5TFhwoSYPHlyLFiwINauXRtbbbVVtG7dOg444IBo2LBhVR5CsRYtWhQTJkyIKVOmxMKFCyOdTsfWW28d7dq1i/322y/q1KlTofYKLiRvbqZMmRIff/xxzJ07N1avXh3NmzePPffcM3bfffdNam/27Nnx4YcfxsyZM2Pp0qVRo0aNaNiwYbRq1So6deoUbdq0qdwDSKAlS5bEqFGjYubMmbFy5cpo3rx5HHTQQdG2bdvq7lq16tatW3z00UdxzDHHxJtvvhkREV988UVcfPHF8fDDD5dat127dnH99dfHJZdcEhER7777btxyyy1ZP3oo8Mc//jE+//zzzPJtt90W2267beUdyGZgzZo1MXr06Jg2bVrMmzcv6tatGzvssEMcfPDB0bRp001qsyq/31avXh3/+c9/YurUqZGfnx877LBD/OQnP4lOnTptcpsbGz9+fHz66acxd+7cSKfTse2220a3bt1ip512qpT2Z8yYER9++GHMnj07vv/+++jUqVOJc2c5zwJANUoDACTA0KFD0xGRebzxxhtZ29evX5/+5S9/mVWmZcuW6cmTJ5d7H0uXLk1ffvnl6aZNm2a1U/BIpVLpgw46KP3RRx+Vq73i+vv++++n99tvv2Lbr1OnTvoPf/hDevXq1SW22b59+0z5/v37l/vY0ul0+q233sra35tvvlmkzKeffpq+7LLL0rvsskuxfSx41KtXL/3rX/86PW/evAr1obwOPvjgzL5at25d4forVqxId+7cudTPTGHffvtt+uabb07vv//+6Ro1apR43Lm5uenjjz8+/emnn5a7/+V5HHzwwUXamDZtWvrPf/5zes8990zn5OSUWLdWrVrps846Kz1jxoxS+1SR/mz8mhf++yvt/XjjjTey2imPdevWpf/xj3+kd9555xL7s9NOO6Uff/zxcrWXTqfT7777bvqQQw5Jp1KpUo9zhx12SF9wwQXpVatWlbvt4nTo0CHT5k033VRq2fPOOy+rD5deemmp5a+++upM2b322qvYMqV9zjc+d5bnMX369Kw2+vfvX+S8s3Tp0vS5556brlOnTrFt9OzZMz1lypRSj21TFH49iutreVTk/NK6detS/07LMmfOnPR2222XdQ6ZOnVqmfXWrVuXPuCAA7K+Hzb+Pnv//ffTubm5mTJHH310hftXHoVf74p+76TTRc8LQ4cOLVe9uXPnps8///x0w4YNSzwfH3PMMeX+nFX199vatWvTN9xwQ3qrrbYqtt299947PWbMmHQ6Xf5zamHff/99+i9/+Uu6RYsWJfb9Jz/5Sfq1114rV3uFP9sF78kHH3yQ7t69e5Fz5x577FGk/o99ngUAihIOAQCJUFo4tHbt2nS/fv2ytrdr167Mi+WFffjhh+ltt922XBdOc3Jy0nfffXeZbW7c36FDh6Zr1apVZvt9+vRJr127ttg2b7jhhqwLWMuWLSv3MZ511llZF9uLs9dee1XoInKrVq3KDEo2xQ8Nh9LpdPr555/P6mvv3r1LLHvppZdW6Lhr166dfuKJJ8rV//I8irvofOKJJ1aoja222io9atSoEvtUkbZ+rHBo/vz56W7dupW7X2eeeWaJfxsF/vGPf5QaphX3WLRoUZl9LU3hwKe0z1k6nR0kRZQc+BQo/FkqKUgq6dyYTldNODR9+vSsoLqkR7NmzdKfffZZma9fRWxp4VA6nU7ffvvtWX2+4IILylVv8uTJ6dq1a2fqHXDAAel169al0+kNQcGuu+6a2da4ceP0N998s0n9K0vhvv9Y4dCLL76YbtCgQbk+r3l5eel///vfZbZZld9v33//ffqII44os82aNWumH3744QqHQ19++WW5/uYKHldccUWZbW4cDg0dOjRds2bNYtvbOByqjvMsAFCUYeUAgERbvXp19O3bN5555pnMuk6dOsVrr70W2223XbnaeO+996Jnz56xfPnyzLo999wzjjrqqGjVqlWsXbs2xo8fH08++WQsXrw41q9fHwMHDoyGDRtG3759y7WPUaNGxXXXXRdr166N1q1bxwknnBDt27ePVCoVEyZMiIcffjiWLVsWERH//ve/44477oiLL764SDsDBgyIP/zhD7F27dpYsWJFPPnkk/Hzn/+8zP2vWLEinnrqqczyWWedVWr5VCoVXbp0iW7dukW7du2icePGsXLlypg0aVI8//zzMWPGjIiImDlzZhxzzDExfvz4ahlurTS9e/eOdu3axZdffhkRG96DtWvXRo0apf8Xum3btnHggQdG586do0mTJrF+/fqYOXNmvPrqq/Hhhx9GxIZ5QM4444xo165d7LXXXkXa2GGHHaJdu3axcuXKmD17dmZ9u3btit1nWcMV7rrrrrHffvvFLrvsEltttVWsXr06pk2bFiNGjIjPPvssIjYMPXfsscfGhAkTolWrVkXaKLzvr776KtauXRsREdtss000aNAgq2yLFi1K7U9l+O677+LAAw+MyZMnZ+33uOOOi44dO0ZeXl5MnTo1nnrqqZg2bVpERDz88MNRp06duOeee4ptc/z48XHeeedl5mfJy8uLo48+Ovbee+9o1qxZpNPpWLRoUXz22WcxevTomDJlSqUcS48ePTJ9euedd2LdunWRm5tbpNycOXOyjjci4pNPPoklS5ZEo0aNipT//vvv47333svaT0U1bNgw896X9b4XKG1Irfz8/Dj22GNjypQpUbt27ejTp09069YtGjVqFN988008+eSTMXHixIj4vyEd33///cjJSe50uWeddVZcccUVmXP8a6+9Vq567du3j2uvvTZ+97vfRUTE6NGj429/+1tccsklcfXVV2f+9iMibrnllth+++0rv/PV4F//+lecfPLJmc9qRET37t3j8MMPjx122CHy8/Pjgw8+iGeeeSZWrlwZq1atihNPPDFee+21OPjgg8tsvyq+3/r37x+vvvpqZrl+/fpx8sknR5cuXSI3NzcmTJgQjz/+eCxevDh+8YtfxAUXXFDu12Pq1KnRvXv3mDNnTmZd+/bto0+fPtGuXbvIycmJzz77LJ544olMmeuvvz7q168fgwYNKtc+3n///bj//vtjzZo1sc0228SJJ54Yu+66a9SqVSumT5+eNXRhdZ1nAYBiVHc6BQDwYyjuzqH8/Px0r169stZ36dIlvWDBgnK3u2TJknSbNm0y9Rs2bJh+9tlniy07d+7cdI8ePTJlt9pqq/Ts2bNLbLtwvwp+YTto0KBih1b56quv0m3bts2U32abbdJr1qwptt3jjjsuU+7AAw8s13E+8MADmTq5ubnpWbNmFVuuR48e6SuuuKLUX+OvXbs2feONN2YNJXPZZZeVqx/lVRl3DqXTRYfwevfdd4std9lll6X79u2bfv/990ttb8SIEVlDBu2zzz6llt+UYdYK9O3bNz1w4MD0xIkTSy03bNiwdF5eXmYfp5xySpltFzecUGmq4s6hE044IVMulUql//SnPxX7t7Fq1ar0JZdcktXuSy+9VGybZ555ZqZM+/bty7yr5NNPP03/4he/SC9durTUcmX59ttvs/r34YcfFlvusccey/o7LHj+/PPPF1u+8OuZm5ubXrx4cbHlNj43lqSi73uBwncOFZzLunbtWuzru3bt2iJ/d88991y591WWLfHOoXQ6nf7pT3+a1e9vv/22XPXWrl2b3nfffTP16tatm3700UezPj+9evXa5H6VR+F+V/WdQzNnzkw3btw4U3b77bdPv/XWW8WWnTZtWrpTp06ZsjvuuGN6xYoVJbZdVd9vw4cPzzq+PffcM/3VV18VKTdnzpx09+7ds/6OyvoMrlmzJr3PPvtkytaqVSv997//PXMHWWFLly5Nn3rqqZmyNWvWTE+YMKHEtgt/tgseP/vZz8o8H1bXeRYAKEo4BAAkwsbh0PPPP58V1ERsGHKnpIunJfn973+fddGlYD6Akixfvjy90047Zer89re/LbHsxhddLr744lLbfv3117PKv/zyy8WWe+GFF7LKlWe+hYILUmVdSFy5cmWZbRW46qqrMm1uvfXW6e+//77cdctSWeHQxp+bu+66q9hyFTnujd+n0uag+iHhUEX6dP/992ddECzrwnN1h0MvvfRSVrmbb765zD707ds3U75r167FlmnZsmWVBBLlUZ55hwqHJmeccUbmeUnDxZVnvqF0+scNhwo+A6Wda1etWpVu165dpvxpp51W7n2VZUsNhzbu94svvljuuv/973+zAuDCj4YNG6a//vrrTe5XefyY4VDhcKNx48Zlzs/07bffpps0aZKpc8cdd5RYtqq+3zp27Jgp26xZs1LPv4sWLcr6QUpZn8G77747q+zTTz9dar/Xrl2b9X1/0kknlVh243CoZ8+exYZOG6vO8ywAkC259+YDAIl29tlnx6hRozLLhx9+eLzyyivFDs1Ukvz8/Lj33nszyxdffHHst99+pdapV69e3HjjjZnlYcOGxZo1a8rcV5MmTWLIkCGlljn00ENjxx13zCx/8MEHxZY78sgjs4b9GjZsWKntTp06Nd5+++3McmnD0NWuXbvUtgq7/PLLo379+hGxYYiwjz/+uNx1fyxt2rTJWv7uu++KLVeR4z700EPjkEMOySy/8sorm9S3slSkT2eddVZm6LA1a9bEf/7znyrpU2W59dZbM8/33nvv+M1vflNmnb/+9a+Z4c4++uijGDduXJEyhYdd2nnnnX94Ryug8JBvhc9NhRWsz8nJiauuuqrc5Tduv7rdeOONpZ5ra9WqFf37988sl3Quqwxt27aNVCpVocebb75ZZf0pSXnPRcXZdddd4+qrry5228033/yjDANZ4MEHH6zw6134fFmaWbNmZQ0Re91115U4FGeBbbfdNq688srMcuHv9I1VxffbmDFjYtKkSZnlq666KrbddtsS223cuHFcf/315epDOp2O2267LbN88sknx4knnlhqndzc3Kzz63PPPRfz5s0r1/5uvfXWcg3/WJ3nWQAgm3AIAEikwhc72rVrFy+88ELUq1evQm28+uqrsWjRoszyRRddVK56xx57bGb+gQULFsT48ePLrHPKKadEnTp1yizXrVu3zPON5yYpkJubGwMGDMgsP/TQQ5mx/4szdOjQzPOmTZtGnz59yuxHedStWzerv5tjONS4ceOs5YULF1ZKu4cddljm+eZw3BtfgN0c+lSSRYsWZc3NUdzcWsVp3rx59OzZM7P8+uuvFylTt27dzPPCc/X8GAqHN2+//XasW7cua3vh+Yb23HPPaN++feyyyy4RETFu3LhYsmRJVvnKmG+oKjRo0KDMC9QR2eey6dOnlytE/1/2Q89Fv/vd74qEDj179oxzzjnnh3Zts/H0009n5hmqX79+uebTi4g488wzM88nTJgQCxYs+MF9Ke/32wsvvJB5XqtWrejXr1+ZbZ900kmx1VZblVlu/PjxWcFTec+VXbp0iV133TUiNvxY4K233qpQnbJU53kWAMgmHAIAEu/LL7+MwYMHV7je6NGjM887d+5c7l9f5+bmRpcuXTLLH374YZl19t1333K1XXhC8cWLF5dY7uyzz45UKhURG35tXdIE5+vXr4+HHnoos/yzn/0satWqVa6+lEfz5s0zz7/55ptKa7eyFPzyu0DBhPA/1OZ43Jtjn4ozZsyYSKfTmeUjjzyy3HX32WefzPPi/u5+8pOfZJ7/+te/juHDh2ftqyoVDm+WLl0aY8eOzdpe3F1ABf+uX7++yAXc9957L1atWhURG8453bt3r/Q+b4ouXbpEjRo1yixX+FyWTqeLhF+VpXXr1tGuXbsKPSpyB0ll+aHnoieffDLrjo2IDQHij/X5LtCgQYMKv96FPwulKfyd3KNHj3K/T82aNcu6M6s838nlUZ5zauF97b333uW6e7lmzZrlCnsLvx6NGjUq887mwso6V26sIm1X53kWAMhW9v/KAQD+B1122WVx2223ZS6eDh48OHJzc+Oaa64pdxsTJkzIPC/4BX95Fb5oNGvWrDLLlzbMTGGF735asWJFieXatGkThx12WIwcOTIiNtwd9NOf/rRIuddeey2rf+X9JfbcuXPj8ccfj7feeismTpwY8+fPj2XLlmV+1V2cqrr4+0NsfAG24I6vkuTn58dzzz0XL7/8ckyYMCFmzZoVy5Yty3zOilPVx7148eJ4+umn4/XXX49PP/005syZE0uXLi31TozN8b0oUPjvrlmzZrH11luXu25Zf3fnnXdeZsiwJUuWxAknnBCtWrWKPn36xCGHHBLdu3ePZs2a/YDel2zbbbeNDh06ZO4OGjVqVOy9996Z7SWFQ3fffXdm+zHHHFNs+Z/85CcVGjKzKm3KuSxiw/msadOmld6fUaNGFRmyrSw9evT40YeWq+i5qLB58+YVe2fr22+/HXfddVdccMEFP7h/5XXCCSeUOZTpxkaNGlWuoeV+6HfyjBkzIqLs7+TK/H6bMmVK5nmnTp3K3d/OnTvH8OHDSy1T+PVo3759uYZ8K1DR/6OUNXxfYdV5ngUAsgmHAIBE6tWrV3Tv3j1OPPHEWL16dURE/OlPf4oaNWrEH/7wh3K1UXjOh6eeeipzJ05FlXaHT4FN+aV6Wb/EPeecczLh0L/+9a9YtGhRkaFqHnjggczzvfbaK3bfffdS21y9enVcc801cfPNN2de1/L6/vvvK1T+x7Dxe9OkSZMSyz788MNx6aWXxvz58yu0j6o67nQ6HbfccktcffXVsXz58s2iT5Wh8N/d/PnzK/Xv7vTTT4833ngj7rvvvsy6mTNnxh133BF33HFHpFKp6Ny5c/Tu3Tv69+8fHTt23KR9l6RHjx5Z4dDvfve7zLaCsCc3NzcOOuigTPmNtxe3vLkMKRexaeeyiLLPZ//rKnIu2tjAgQMzfze5ubnRpk2b+PLLLyNiw9w4Rx99dLRu3brS+lpdCp8bbrrpprjppps2qZ2SvpOr4vut8L4qEoiUp2zh1+PDDz+s0v+jNGjQoNztVfd5FgD4P4aVAwAS6+ijj46nnnoqM0l9xIbJoIcMGVKu+pV1d0V+fn6ltFNRxx9/fOaui1WrVsVjjz2WtX3RokXx3HPPZZbLumto3bp1cdJJJ8WQIUOKXDjLzc2NbbbZJlq1apU1XFDhC0qb48Xfgl+SFyjpzoW//OUv0a9fv2KDoSZNmkSLFi2yjnubbbbJbK+q477gggvi0ksvLRIMpVKpaNq0abRs2TKrT4WDwc3xvShQ1X939957bzz88MOx0047FdmWTqfj008/jRtuuCF23XXX+NnPfpY179gPVTjEeeeddzLzDm0831DBXSPbbLNN5g6JTz75JHMRd3Odb4hNV95z0caefPLJeOaZZzLLl156aTz11FOZof2WL18e5513XqX1szpV5bmhqr7fCt/hW555BQsUnrenJD/m/1HKM1RkYdV5ngUA/o87hwCAROvTp0888cQTccopp2SGhLniiiuiRo0aWb/aL07hizMNGzbc5GFQCgcFP6ZatWrFmWeeGbfeemtEbBhabuDAgZntjz76aGY4tNq1a0ffvn1Lbe/vf/97PP/885nlPfbYI371q19Fjx49ok2bNpGbm1ukTv/+/bPmNNrcvPvuu1nLheeKKjBhwoS44oorMsvNmzePiy++OI488sjYddddIy8vr0idoUOHlnuIvk0xYsSIzHBjERE77rhjXHzxxXH44YfHzjvvnBWIFrj66qvjz3/+c5X1qbIU/rurWbNmtGrVapPaKW2OsDPOOCN+9rOfxbvvvhuvvPJKvPXWW/Hee+9l/fo/nU7Ho48+GmPHjo133303GjduvEn9KGzjeYc+/vjj2GeffUq9C6hHjx7x+eefZ+Yd6tOnz2Y73xCbrjznoo0tWLAgLrzwwsxyhw4d4k9/+lPUrl07fv/738d1110XERGvvPJKDBs2LAYMGFCpff6x1a1bN5YuXRoRG0L5je+ELa/i6lXV91u9evUyfV65cmW5+1iewKbwubJOnTrlnrtpY5taryzVdZ4FAP6PcAgASLzjjz8+HnvssTj99NMzAdFll10WNWrUiF//+tcl1is810mfPn3i4YcfrvK+VrZzzjknEw599NFHMXHixOjcuXNEbAgwChx//PFlXpT529/+lnl++OGHx4gRI6JWrVql1inPcDXVZf369fH6669nlhs0aFDsBdk777wzc4fHtttuGx9//HGZF9Oq+rgLvxedO3eO0aNHlzlHyeb8XhRW+O+uefPmMXXq1CrZTyqViv333z/233//iNhwd91bb70VTz75ZDzyyCOZC7mTJk2KP/3pT3HLLbf84H0WN+/QxuHQxnOvHHLIIVnzDvXp02eznW+ITbN8+fKscGiXXXYp148RLrjggszdjDk5OfHAAw9khvW76qqr4tlnn43PP/88IiJ+85vfxJFHHlnuOaE2R1tvvXUmaBk4cGBce+21ldZ2VX2/NW7cONPnigxJWp6yhc+Ve+21V7z99tvlbv/HUh3nWQDg/xhWDgAgIk466aR45JFHsn79+5vf/CbrgtDGOnTokHn+zTffVGn/qkqnTp2iW7dumeWCQGjChAnx8ccfZ9aXdZfLN998kzWx9uDBg8u8cBYRMX369Ip2+UczYsSImDZtWmb5kEMOKfbX4f/5z38yzy+55JJy/cq6Ko97/fr1WeHAH/7wh3JNXr85vxeFFf67mz9/fqxZs+ZH2W9eXl707Nkz7rvvvvjvf/+bNWH7xkMy/hDFzSNUeL6hje8CKq38xtvZMg0dOjRreMiePXuWWefZZ5+NJ598MrN88cUXZy7AR2z4PN9///2Rk7PhksCiRYuy7hzdElXVd3JVfr+1b98+8/y///1vufs0ceLEMstsif9H+bHOswDABsIhAID/3ymnnBIPP/xwVgBw8cUXx5133lls+YMPPjjz/P33388M47SlOeecczLP//nPf8batWuz7hpq3bp1HHbYYaW2MXv27KzlPfbYo8z9zp8/v0IXw35M+fn5WUPFRUSJwwwWPvbyHHdExBtvvFGuchsP/7Z+/foy63z33XdZc2KUp0+rV6+O0aNHl6tP1a3w392qVauy5tb5sbRt2zYGDRqUWZ47d24sXLiwUtreeN6hWbNmZe4k6tKlS5GJ35s1axa77rprRESMHz8+5syZU2XzDRX+PJbns8gPN2/evKx58GrUqBGXXHJJqXUWLlyYFfTstNNOmSHkCttvv/3ioosuyiwPHz48nnrqqR/e6WpS+Nzw1ltvVVq7Vfn9tvfee2eef/jhh+WaJ2jNmjVZAXBJCr8e06dPj6+//rrMOpuTqjzPAgAbCIcAAAo5/fTTY9iwYZlfU0dE/OpXv4q///3vRcoeeeSRmQu1+fn58cADD/xo/axMp556atSvXz8iNlyIHD58eDzyyCOZ7QMGDIhUKlVqGxtPtl14zoCS3HXXXZvlBeY1a9bE2WefnfXL7B49esSBBx5YbPnCx16e437zzTfL9avviMi8LwUKhh8qzaa8F4899tgWc9Ft2223zXov7rjjjmrpR+Ff/EdEZkjKH6pwmLNs2bL461//mlneeEi5jeusX78+/vKXv1TZfEOFP4/l+SzywyxbtixOOeWU+PbbbzPrzjjjjGjbtm2p9X71q1/F3LlzI2LDsF33339/1KlTp9iy1113Xey4446Z5QsvvDC+++67Suj9j+/EE0/MfHd/+eWX8dJLL1VKu1X5/Xb00Udnnq9evTr++c9/llnn2WefjUWLFpVZbu+99442bdpklqvrXPlDVNV5FgDYQDgEALCRM844Ix544IHMRaZ0Oh0DBw6M++67L6tc48aN47zzzsssX3HFFZn5G8qrPBeZqlr9+vXj1FNPzSz/6le/ysxnkEql4qyzziqzjZYtW2YtjxgxotTyn376adxwww2b0NuqNWnSpDjooIPi8ccfz6xr3rx5DBs2rMQ6hY+9rONetmxZnH/++eXuT+vWrbOWyxMqbb311lkTkZfVp9mzZ5d4V9Tm6vLLL888f/LJJys83NC6deuKXGTMz8+v0Jwfn376aeZ53bp1o2nTphXqQ0kK5h0qUDiYLukuoMLrC5ev7PmGCn8eyxtwsmnee++92HvvvePNN9/MrOvYsWOZc678+9//jkcffTSzfMEFF8RBBx1UYvm6devGP/7xj8wPAObNm1fmnUmbq5133jlOOOGEzPL555+fCcnKq7jv5Kr8ftt///2jY8eOmeU///nPpfZ5yZIlRe5qLUlubm789re/zSzfeuutWZ+n8qjs/6NsLudZAGAD4RAAQDH69+8f9913X+aCWTqdjvPOOy9ruLWIDfO5tGvXLiI2TD594IEHxhNPPFHmL4bHjx8fv/nNb2KfffapmgOooMJDyxW+MHXooYcWCSiKs91220WnTp0yy5deemmJQ+r85z//icMOOyy+//77rDu0qsP69evjv//9b9x///1x1FFHxa677po1JFf9+vXjueeeK/U1KDz/x9ChQ+OJJ54otty0adPisMMOi0mTJpX7uBs3bpz1y+lrrrmmzAtrubm5WXeYDBkypMQLgp988kkcdNBBMX/+/Gp/Lyqid+/eceKJJ2aWzzzzzPjTn/4UK1asKLXerFmz4uabb4527drFrFmzsrbNmzcv2rRpExdffHF88sknpbYzevTouP766zPLRx99dKW+foXDnoIJ2WvUqFHi3WvFld94fWXYd999M8+feOKJSh26K+nWrFkTY8eOjbvuuisOPPDA2G+//TLDCUZsCKlfeOGFaNy4cYltLFq0KCt8btu2bblCikMOOSR+8YtfZJb/+c9/xosvvrhpB1LN/t//+3/RpEmTiIiYOXNm7LPPPvHaa6+VWmf9+vUxZsyYOOecc+K4444rsr2qv98KDxs4b9686N27d7FDwM2bNy+OPfbYmDZtWrnbPvfcczPzCq5evTp69eoVd955Z5lztX3xxRdxzTXXRKtWrcq1n/LanM6zAEBEjeruAADA5urnP/95rFu3Ls4777xIp9ORTqfjnHPOidzc3OjXr19ERDRq1CieeeaZOOSQQ2LRokWxcOHCOO200+KKK66II444InbZZZdo2LBhrFy5MhYsWBATJ06M999/P7766quIKHpnSHXp1q1bdO7cucjdAD//+c/L3cbvf//7zOsyd+7c2GuvveLEE0+M/fbbL+rVqxezZ8+OV199NXNBebfddouOHTtW+RwX33zzTey0005Z69auXRuLFy+OpUuXFhkyqMBee+0VDz30UGY+l5L8+te/jvvuuy9Wr14d69ati9NOOy3uu++++OlPfxrNmjWLRYsWxZgxY+L555+PVatWRf369WPgwIHxl7/8pVz9HzBgQOaX4q+//npsu+220aZNm6y5Z7p27Rr/+Mc/MsuXXXZZ5tftK1asiEMPPTSOOeaY6NGjRzRu3Djmz58fb7zxRrzyyiuxfv362H777aNPnz7FDp+4uXrggQdi6tSpMX78+Fi3bl1cc801cdttt8WRRx4ZXbp0iSZNmsS6deti0aJFMXny5Pj4449j/PjxpbaZn58ff/vb3+Jvf/tbtGvXLvbff//o1KlTNG3aNFKpVMyePTveeuutGDlyZOZzU7t27bj22msr9dh69OgR99xzT9a6vfbaq8h8QwUK5h367LPPirRTmfr27RvXXHNNrF69OpYvXx4HH3xwNGvWLJo3b541V9uLL74Y22+/faXu+3/B+++/X+RctHr16li8eHEsX768xHPR4YcfHkOHDo0WLVqU2v4ll1ySGYIulUrFP/7xj6hXr165+nbTTTfFiy++mAlNzzvvvPjvf/8bDRs2LFf9zUXr1q3jsccei2OPPTa+//77mDlzZhxxxBHRqVOn6NmzZ7Rr1y7q168fK1asiHnz5sWECRPi3XffzfwoovA8PYVV5ffbcccdF6eeemrmhwUff/xx7LrrrnHqqadGly5dIicnJz799NN47LHHYtGiRVG7du244IIL4uabby6z7Zo1a8ZTTz0VBxxwQMycOTNWrlwZF154YVx33XVx5JFHxm677RZbbbVVrFq1KhYuXBifffZZfPjhh1nBZGXbXM6zAEBEpAEAEmDo0KHpiMg83njjjXLXvfvuu9OpVCpTNycnJ/3Pf/4zq8zkyZPTHTp0yNpHeR477rhjifvdlP5effXVmToHH3xwuY8xnU6nb7nllqx9Nm7cOL1y5coKtfHzn/+83Mf9xRdfpPv3759Z179//wrtqzQHH3xwhd+Lgsc+++yTfuSRR9Lr1q0r9/6GDh2azsnJKbPt+vXrp//9739nfR5bt25daturVq1KH3rooaW2W9x7/ac//alcx9usWbP0e++9V6HPTuvWrTNlhw4dWq7XpzzH+8Ybb2T1rSzLli1L9+nTZ5Pe56+++iqrrenTp1e4jQYNGqRfffXVMvtZUd9++22Rff3+978vtc7AgQOzyufm5qYXL15crv1V5Fzz97//PZ2bm1vq6zJ9+vSsOpvyd77x+7Fxm5uq8Od8U9stfH4p6++38N9KRR45OTnpww47LP3CCy+Uq08jRozIqn/eeedV+LheeOGFH9xGcQq3uSnn+Y3PC+U557z77rvp7bbbrsKv+6GHHlpim1X5/fb999+nDz/88DLbrlWrVvqf//xnhb5D0ul0es6cOen99ttvkz6HJano90A6vXmdZwGAdNo9uQAAZTj//PPj9ttvzyyvX78++vfvnzUvTfv27WPChAlx1113FZlAeWM1a9aMgw46KP76179mDWFW3c4444zMMHoREaeffnrUrl27Qm384x//iFtuuSW23nrrYrfXr18/zjvvvBg3blyRX9D/mGrWrBkNGzaMHXfcMQ466KA499xz47777ovp06fH+++/H3379q3Q8DUDBgyIF198MWvuiMJyc3PjyCOPjI8//jiOOeaYCvW1Vq1a8eqrr8YjjzwSxx13XLRp0ybq1auX9V4V549//GP885//LDJfRoG8vLw49dRTY/z48VnDhW1JCob9e/HFF6N79+5lvmedO3eOyy+/PD7//PMiwyVtt912ce+998bRRx9d5jw99erVi7POOis+++yzrGEFK8vG8w5FlH0XUOGhBCMqf76hAuedd158/PHHMXDgwPjJT34SjRs3zrpriIqpVatWNG7cOHbeeec47LDD4sILL4yHH344vvnmmxg5cmT07t27zDaWLFkS5557bma5ZcuWcdNNN1W4L717946f/exnmeV777033njjjQq3szno1q1bfPHFF3HdddeVeA4sUKdOnTjiiCPi3nvvjeHDh5dYriq/3/Ly8uLll1+OIUOGlDh04N577x2jRo3Keo/Kq3nz5vHOO+/Eo48+GnvuuWepZXNycmLvvfeOa6+9NqZPn17hfZVmczrPAgARqXS6hHvXAQDYZDNmzIgPPvgg5s2bF4sXL466devG1ltvHR06dIjddtut3EP9/JjeeuutrCF1Pvroo9hrr702qa3vv/8+3nnnnfjss89i+fLl0bRp02jZsmUcfPDBUbdu3crq8mYnnU7Hxx9/HB9//HF899130bBhw9h+++1j//33j2233bZa+rR27dp47733Yvz48bFkyZLYaqutYocddoiDDjqo1PlLtkSLFi2Kd955J2bPnh3fffdd1KhRIxo3bhw77bRT7LbbbtGsWbNytbN+/fr4/PPPY/LkyTFr1qxYtmxZ1KhRI7baaqvYZZddYq+99vqf/hzD/5rPP/88xo0bF/Pnz49ly5ZFvXr1onnz5tGxY8fo1KlT5OXllbutqv5+W7VqVbz++uvx5ZdfxsqVK2P77bePPffcM2veox9qzpw5MWbMmJgzZ04sWrQo8vLyokmTJrHzzjvHbrvt9qN8NzjPAkD1Ew4BABAREWeddVYMGzYsIiJ23333MudnAQAAALZMhpUDACAWL16cNXH2L37xi2rsDQAAAFCVhEMAAMStt94aK1asiIgN4/yfeeaZ1dwjAAAAoKoIhwAAEu7pp5+OG264IbP8y1/+skomsgcAAAA2D+YcAgBImI8++ijOOeecSKfTMWvWrFi4cGFmW7NmzWLSpEnRpEmTauwhAAAAUJVqVHcH2DTr16+P2bNnR4MGDSKVSlV3dwCALcjcuXNj/PjxRdbn5eXFvffeGzVq1IilS5dWQ88AAACATZVOp2PZsmWx/fbbR05O6QPHuXNoCzVr1qxo2bJldXcDAAAAAADYjHz99dfRokWLUsu4c2gL1aBBg4jY8CY3bNiwmnsDAAAAAABUp6VLl0bLli0z+UFphENbqIKh5Bo2bCgcAgAAAAAAIiLKNRVN6YPOAQAAAAAA8D9FOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgNaq7A8Cma3P5iOruArAZmHFD7+ruAgAAAABbEHcOAQAAAAAAJIhwCAAAAAAAIEGEQwAAAAAAAAkiHAIAAAAAAEgQ4RAAAAAAAECCCIcAAAAAAAASRDgEAAAAAACQIMIhAAAAAACABBEOAQAAAAAAJIhwCAAAAAAAIEGEQwAAAAAAAAkiHAIAAAAAAEgQ4RAAAAAAAECCCIcAAAAAAAASRDgEAAAAAACQIMIhAAAAAACABBEOAQAAAAAAJIhwCAAAAAAAIEGEQwAAAAAAAAkiHAIAAAAAAEiQLTocmjx5ctx+++0xYMCA2G233aJGjRqRSqVi8ODBFWrnrrvuilQqFalUKs4555xSy3788cdx8sknR/PmzaN27drRtm3b+NWvfhXz5s0rtd7cuXPjwgsvjLZt20ZeXl40b948Tj755Bg7dmyF+goAAAAAAPBDbNHh0N133x0XXXRRPPjggzFx4sRYt25dhduYNm1aXHbZZZFKpcos+/TTT0e3bt3i6aefjtatW8exxx4bOTk5cccdd8Tuu+8eU6dOLbbelClTYvfdd48777wzcnJy4rjjjovWrVvH008/Hfvuu28MHz68wv0GAAAAAADYFFt0ONS5c+f47W9/G4888kh8/vnnceaZZ1ao/vr162PAgAGRSqWiX79+pZadPXt29O/fP9auXRv33HNPfPDBB/HEE0/ElClT4owzzoi5c+dG3759I51OZ9VLp9Nx2mmnxbx58+LMM8+MKVOmxBNPPBEffPBB3HPPPbF27dro169fzJkzp8LHDwAAAAAAUFFbdDh0zjnnxE033RR9+/aNjh07Rk5OxQ7ntttui7fffjtuvPHGaNOmTallb7311sjPz4/DDz88zj333Mz63NzcuPvuu6NRo0bx4YcfxquvvppV76WXXopx48ZF48aN46677orc3NzMtnPPPTcOO+ywWL58edx2220V6jsAAAAAAMCm2KLDoR9i8uTJceWVV8bBBx8cv/zlL8ssXzD0W9++fYtsq1+/fvTp0yciIp599tli6/Xp0yfq169fpG5BexvXAwAAAAAAqAqJDIfWrVsX/fv3j1QqFffff3+Z8w0tW7YsM59Q165diy1TsH7cuHFZ6wuWy6r3xRdfxIoVK8p/EAAAAAAAAJugRnV3oDrcdNNN8f7778ctt9wS7dq1K7P8jBkzMs9btWpVbJmWLVtGRMT06dOz1hcsl1UvnU7HjBkzolOnTsWWW7VqVaxatSqzvHTp0jL7DQAAAAAAsLHE3Tk0ceLEuPrqq2P//fePiy66qFx1li1blnler169YssUDBm3cWhTULesesXVLWzIkCHRqFGjzKMgVAIAAAAAAKiIRIVDa9eujf79+0dOTk488MADkZOz5Rz+oEGDYsmSJZnH119/Xd1dAgAAAAAAtkCJGlbuuuuui7Fjx8aNN94YHTp0KHe9Bg0aZJ6vWLEiGjVqVKTM8uXLIyKiYcOGReouXLiwxPmECuoVV7ewvLy8yMvLK3efAQAAAAAAipOocGj48OEREfH888/Hiy++mLWtYF6hESNGRI8ePSIiYtSoURER0bp160y5mTNnxm677Vak7YI7edq0aZO1vk2bNrFw4cKYOXNmsX0qqJdKpbL2AwAAAAAAUBUSFQ4VeOedd0rcNmfOnJgzZ07WuoYNG8ZOO+0UU6dOjY8++qjYcOijjz6KiIguXbpkre/SpUuMHTs2s72kejvvvHPW/EMAAAAAAABVYcuZdKcSfPLJJ5FOp4t9XH311RERcfbZZ2fWFXb88cdHRMSjjz5apN3ly5fH888/HxERJ5xwQrH1/v3vfxc7tFxBexvXAwAAAAAAqAqJCod+iEsuuSTq1q0bI0eOjPvuuy+zft26dTFw4MBYvHhx7L333nHEEUdk1evVq1fsueeesXjx4hg4cGCsW7cus+3ee++N119/PerXrx8XX3zxj3YsAAAAAABAcm3Rw8qNHTs2Bg4cmFn+8ssvIyLinnvuiRdeeCGzfvjw4bHddtv9oH1tv/32MWzYsDj99NPj3HPPjfvvvz/atGkTH374YUybNi2aN28ejz76aKRSqax6qVQqHnvssejevXs89NBD8c4778Tee+8d06dPjw8++CBq1KgRDz30UGy77bY/qH8AAAAAAADlsUWHQ0uXLo3333+/yPpZs2bFrFmzMsurVq2qlP2dfPLJseOOO8b1118fb7/9dowbNy622267uOCCC+Kqq66K5s2bF1uvQ4cOMWHChBg8eHC88MILMXz48GjUqFGccMIJceWVVxaZpwgAAAAAAKCqpNIbT67DFmHp0qXRqFGjWLJkSTRs2LC6u0M1aXP5iOruArAZmHFD7+ruAgAAAADVrCK5gTmHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACTIFh0OTZ48OW6//fYYMGBA7LbbblGjRo1IpVIxePDgYsuvX78+xowZE3/84x/jwAMPjK233jpq1qwZTZs2jZ49e8YjjzwS6XS61H1+/PHHcfLJJ0fz5s2jdu3a0bZt2/jVr34V8+bNK7Xe3Llz48ILL4y2bdtGXl5eNG/ePE4++eQYO3bsJh8/AAAAAABARdWo7g78EHfffXfcdttt5S4/bdq0OOCAAyIiokmTJtG1a9fYaqutYtq0aTFy5MgYOXJkPP744/HMM89ErVq1itR/+umn4/TTT4+1a9fG3nvvHW3bto2PPvoo7rjjjnjqqafinXfeiZ122qlIvSlTpkT37t1j3rx5seOOO8Zxxx0X06dPj6effjr+9a9/xZNPPhnHH3/8pr8QAAAAAAAA5bRF3znUuXPn+O1vfxuPPPJIfP7553HmmWeWWj6VSsWhhx4aL730UsybNy9eeeWVePzxx+ODDz6IUaNGRb169eKFF16IG264oUjd2bNnR//+/WPt2rVxzz33xAcffBBPPPFETJkyJc4444yYO3du9O3bt8idR+l0Ok477bSYN29enHnmmTFlypR44okn4oMPPoh77rkn1q5dG/369Ys5c+ZU6msDAAAAAABQnC06HDrnnHPipptuir59+0bHjh0jJ6f0w2nXrl28/vrrceSRR0Zubm7WtoMPPjguv/zyiIh46KGHitS99dZbIz8/Pw4//PA499xzM+tzc3Pj7rvvjkaNGsWHH34Yr776ala9l156KcaNGxeNGzeOu+66K2u/5557bhx22GGxfPnyCt0BBQAAAAAAsKm26HCosu25554REfH1118X2TZ8+PCIiOjbt2+RbfXr148+ffpERMSzzz5bbL0+ffpE/fr1i9QtaG/jegAAAAAAAFVBOFTIF198ERER2223Xdb6ZcuWxdSpUyMiomvXrsXWLVg/bty4rPUFy2XV++KLL2LFihWb2HMAAAAAAIDyEQ79//Lz8+Nvf/tbRESceOKJWdtmzJiRed6qVati67ds2TIiIqZPn561vmC5rHrpdDprPxtbtWpVLF26NOsBAAAAAABQUcKh/9/AgQNj+vTpsf3228cVV1yRtW3ZsmWZ5/Xq1Su2fsGQcRuHNgV1y6pXXN3ChgwZEo0aNco8CkIlAAAAAACAihAORcS1114bDz74YNSuXTuefPLJ2Hrrrau7S0UMGjQolixZknkUNy8SAAAAAABAWWpUdweq21//+tf44x//GHl5eTF8+PA44IADipRp0KBB5vmKFSuiUaNGRcosX748IiIaNmxYpO7ChQtLnE+ooF5xdQvLy8uLvLy80g8GAAAAAACgDIm+c+j222+PSy+9NGrVqhXPPPNMHHnkkcWWa926deb5zJkziy1TcCdPmzZtstYXLJdVL5VKZe0HAAAAAACgKiQ2HLrzzjvjoosuygRDvXv3LrFsw4YNY6eddoqIiI8++qjYMgXru3TpkrW+YLmsejvvvHPW/EMAAAAAAABVIZHh0N///ve48MILM8HQ0UcfXWad448/PiIiHn300SLbli9fHs8//3xERJxwwgnF1vv3v/9d7NByBe1tXA8AAAAAAKAqJC4cuu+++2LgwIEVCoYiIi655JKoW7dujBw5Mu67777M+nXr1sXAgQNj8eLFsffee8cRRxyRVa9Xr16x5557xuLFi2PgwIGxbt26zLZ77703Xn/99ahfv35cfPHFlXOAAAAAAAAApUil0+l0dXdiU40dOzYGDhyYWf7yyy9jwYIF0aJFi9hhhx0y64cPHx7bbbddfPLJJ9GlS5dIp9PRsWPH2HfffUtse9iwYUXWPfXUU3H66afHunXrYt999402bdrEhx9+GNOmTYvmzZvHO++8kxl+rrDJkydH9+7dY/78+bHjjjvG3nvvHdOnT48PPvggatSoEU8++WTmDqPyWrp0aTRq1CiWLFkSDRs2rFBd/ne0uXxEdXcB2AzMuKHkoVEBAAAASIaK5AZbdDg0atSoOOSQQ8osN3369GjTpk25y0dElPSyfPzxx3H99dfH22+/HUuWLIntttsujj766LjqqquiefPmJbY3Z86cGDx4cLzwwgvx7bffRqNGjaJ79+5x5ZVXFpmnqDyEQ0QIh4ANhEMAAAAAJCYcSjLhEBHCIWAD4RAAAAAAFckNEjfnEAAAAAAAQJIJhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkiHAIAAAAAAAgQYRDAAAAAAAACSIcAgAAAAAASBDhEAAAAAAAQIIIhwAAAAAAABJEOAQAAAAAAJAgwiEAAAAAAIAE2aLDocmTJ8ftt98eAwYMiN122y1q1KgRqVQqBg8eXGbdkSNHxlFHHRVNmzaNOnXqRMeOHePKK6+M5cuXl1pv6tSpMWDAgGjRokXk5eVFixYtYsCAATFt2rRS6y1btiyuuOKK6NChQ9SpUyeaNm0avXv3jv/85z8VOmYAAAAAAIAfYosOh+6+++646KKL4sEHH4yJEyfGunXrylXvlltuiZ49e8bLL78cnTp1imOOOSaWLFkS119/fXTt2jUWLFhQbL3Ro0fHHnvsEQ8++GA0btw4jj/++GjcuHE8+OCDsfvuu8d7771XbL158+ZF165dY8iQIbFs2bI45phjolOnTvHSSy/F4YcfHrfffvsmvwYAAAAAAAAVsUWHQ507d47f/va38cgjj8Tnn38eZ555Zpl1xo0bF5deemnk5ubGiBEj4s0334wnn3wyvvzyyzjssMNi8uTJcf755xepl5+fH6ecckrk5+fHoEGDYuLEifH444/HxIkTY9CgQbFixYo45ZRTYuXKlUXqnnvuuTFlypQ47LDDYurUqfHkk0/Gm2++GS+88ELk5OTEJZdcEhMmTKiU1wQAAAAAAKA0W3Q4dM4558RNN90Uffv2jY4dO0ZOTtmHM2TIkEin03HWWWdFr169Muvr1q0b999/f+Tk5MQzzzwTkyZNyqo3bNiwmD17drRv377IsHWDBw+O9u3bx9dffx0PPfRQ1rbPPvssnnvuucjNzY37778/6tatm9l21FFHxYABA2L9+vUxZMiQTXkJAAAAAAAAKmSLDocqavXq1TFixIiIiOjbt2+R7a1bt44DDjggIiKGDx+eta1g+bTTTisSQuXk5MSpp54aERHPPvtssfUOOOCAaN26dZF9FvTj+eefjzVr1lT4mAAAAAAAACoiUeHQlClTIj8/PyIiunbtWmyZgvXjxo3LWl+wXFX1VqxYEV988UWZxwAAAAAAAPBDJCocmj59ekRENG7cOBo0aFBsmZYtW2aVjYhYtmxZfPfddxER0apVq1LrzZ8/P1asWFFknyXVa9iwYTRs2LDIPje2atWqWLp0adYDAAAAAACgohIVDi1btiwiIurVq1dimfr160dEZIUvBfVKq1tQr6S6Fd3nxoYMGRKNGjXKPArCKAAAAAAAgIpIVDi0JRs0aFAsWbIk8/j666+ru0sAAAAAAMAWqEZ1d+DHVDCUXOFh3za2fPnyiIjMUG+F65VWt6BeSXUrus+N5eXlRV5eXonbAQAAAAAAyiNRdw61adMmIiIWL16cNVRcYQV35BSUjdgQ8DRp0iQiImbOnFlqvaZNm2YNIVfQTkn1Cs8fVHifAAAAAAAAVSFR4VCHDh2ibt26ERHx0UcfFVumYH2XLl2y1hcsV1W9evXqRfv27cs8BgAAAAAAgB8iUeFQrVq1onfv3hER8eijjxbZ/tVXX8WYMWMiIuL444/P2law/Pjjj8f69euztq1fvz6eeOKJiIg44YQTsrYdd9xxERExevToYu8eKujHMcccEzVr1qzoIQEAAAAAAFRIosKhiIjLL788UqlUDB06NF5++eXM+vz8/Dj77LNj3bp1ceKJJ0bHjh2z6g0YMCC23377mDJlSlx11VVZ26666qqYMmVKtGjRIvr165e1rVOnTnHsscfGunXr4uyzz46VK1dmtr300ksxbNiwyMnJiUGDBlXB0QIAAAAAAGRLpdPpdHV3YlONHTs2Bg4cmFn+8ssvY8GCBdGiRYvYYYcdMuuHDx8e2223XWb5lltuid/85jeRSqXi4IMPjm222Sbefvvt+Pbbb6NDhw7xzjvvRNOmTYvsb/To0XHEEUdEfn5+dO7cOTp37hwTJ06MiRMnRr169WLkyJHRrVu3IvXmzZsXBx54YHzxxRex3XbbRffu3WPevHnx5ptvRjqdjttuuy0uuuiiCh370qVLo1GjRrFkyZJo2LBhheryv6PN5SOquwvAZmDGDb2ruwsAAAAAVLOK5AZbdDg0atSoOOSQQ8osN3369GjTpk3WupEjR8bNN98cH3zwQaxYsSJatWoVJ510UgwaNCgaNGhQYltTp06Na6+9NkaOHBnz58+PZs2axeGHHx5//OMfo127diXWW7p0aQwZMiSeeeaZmDlzZtSrVy/22Wef+O1vfxuHHXZYuY+5cHvCIYRDQIRwCAAAAIAEhUNJJhwiQjgEbCAcAgAAAKAiuUHi5hwCAAAAAABIMuEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIkBo/9g7nzZsX7733Xqxduzb22GOPaNeu3Y/dBQAAAAAAgMSqtHBo4cKFMWzYsIiI6N27d3To0KFImWuvvTauu+66WLNmTWbdqaeeGg888EDUrl27sroCAAAAAABACSptWLknnngifvvb38YVV1wRTZs2LbL9kUceiauvvjrWrFkT6XQ683jiiSdiwIABldWNcps5c2ZceOGF0aFDh6hTp07Url072rZtG/3794/x48eXWG/kyJFx1FFHRdOmTaNOnTrRsWPHuPLKK2P58uWl7m/q1KkxYMCAaNGiReTl5UWLFi1iwIABMW3atMo+NAAAAAAAgBJVWjj0xhtvRERE9+7dY+utty6y/Y9//GNERKTT6Tj22GPjkksuiZYtW0Y6nY6nnnoq3n777crqSpnef//96Ny5c9x5552xYsWKOOKII+Koo46KVCoVDz30UHTt2jWeeuqpIvVuueWW6NmzZ7z88svRqVOnOOaYY2LJkiVx/fXXR9euXWPBggXF7m/06NGxxx57xIMPPhiNGzeO448/Pho3bhwPPvhg7L777vHee+9V9SEDAAAAAABERCWGQ1OmTIlUKhX77bdfkW1jxoyJ6dOnRyqVisGDB8fw4cPjr3/9a3z44Yex1VZbRUTEww8/XFldKdO5554by5Yti3PPPTemT58ezz33XDz77LMxderU+MMf/hBr166Nc889N77//vtMnXHjxsWll14aubm5MWLEiHjzzTfjySefjC+//DIOO+ywmDx5cpx//vlF9pWfnx+nnHJK5Ofnx6BBg2LixInx+OOPx8SJE2PQoEGxYsWKOOWUU2LlypU/2vEDAAAAAADJVWnhUMFdMzvvvHORbSNHjoyIiLy8vLj44osz67fZZps4/fTTI51O/2h3z3z33XcxYcKEiIgYPHhw1KxZM7MtJycnrrnmmqhTp04sXrw4Pv/888y2IUOGRDqdjrPOOit69eqVWV+3bt24//77IycnJ5555pmYNGlS1v6GDRsWs2fPjvbt28fgwYOztg0ePDjat28fX3/9dTz00ENVcbgAAAAAAABZKi0c+u677yIiol69ekW2jR49OiI2DDm38fbdd989IjbMAfRjyMvLK3fZgrmTVq9eHSNGjIiIiL59+xYp17p16zjggAMiImL48OFZ2wqWTzvttMjJyX65c3Jy4tRTT42IiGeffbbc/QIAAAAAANhUlRYOpVKpiIhYtGhR1vr169fH+++/H6lUKrp3716kXsH8RPn5+ZXVlVLVr18/048//OEPsWbNmqy+XnPNNbFy5cro1atXtGzZMiI2DJlX0L+uXbsW227B+nHjxmWtL1iuaD0AAAAAAICqUKOyGtpmm23i66+/ji+++CJr/XvvvRdLly6NVCoV3bp1K1Jv+fLlERFRp06dyupKme6777446qij4t57740RI0ZE165dIzc3N8aNGxfffPNNnHnmmXHHHXdkyk+fPj0iIho3bhwNGjQots2CIKmgbETEsmXLMndUtWrVqtR68+fPjxUrVhR751VExKpVq2LVqlWZ5aVLl5b3cAEAAAAAADIq7c6hPffcM9LpdDz++OOxevXqzPr77rsvIiJq1aqVGXqtsGnTpkVExPbbb19ZXSlThw4d4t13340jjjgivvnmm3juuefi2WefjenTp8dOO+0UPXr0iIYNG2bKL1u2LCKKHzKvQP369SMiO7QpqFda3YJ6G9fd2JAhQ6JRo0aZR0GoBAAAAAAAUBGVFg6dfPLJERHx9ddfx2GHHRZ///vf4xe/+EU8+OCDkUqlok+fPsXeHfTee+9FKpWKXXbZpbK6UqbRo0fHbrvtFhMnToxHH3005syZEwsXLoznn38+1qxZE2effXacffbZP1p/ymPQoEGxZMmSzOPrr7+u7i4BAAAAAABboEobVu7000+P22+/Pd5///0YM2ZMjBkzJrMtLy8vrr766iJ1Fi9eHKNGjYqIiH333beyulKqxYsXx/HHHx8LFiyId999N2u/Rx99dOy6666x2267xQMPPBBnnHFGHHLIIZmh5FasWFFiuwXD4xW+46jwEHQl1S2ot3HdjeXl5UVeXl4ZRwcAAAAAAFC6SrtzKJVKxYgRI+K4446LVCoV6XQ60ul07LDDDvHMM8/ErrvuWqTOsGHDYs2aNRERcfjhh1dWV0o1YsSImD9/fuy4447FBlKF148cOTIiItq0aRMRG4KlwkPFFVZwJ09B2YgN4VCTJk0iImLmzJml1mvatGmpw9YBAAAAAABUhkoLhyIimjRpEs8++2x8++238e6778aECRPiq6++il69ehVbftddd42hQ4fG0KFDY6+99qrMrpSoIKQp7S6dRo0aRUTEwoULI2LDHEV169aNiIiPPvqo2DoF67t06ZK1vmC5ovUAAAAAAACqQqWGQwWaNWsW++67b3Tu3DlyckrexRFHHBH9+/ePfv36VUU3irXDDjtERMSkSZNiyZIlRbavWbMmxo4dGxERbdu2jYiIWrVqRe/evSMi4tFHHy1S56uvvsoMo3f88cdnbStYfvzxx2P9+vVZ29avXx9PPPFERESccMIJm3xMAAAAAAAA5VUl4dDmrFevXlGvXr1YuXJl/OIXv8ia82f16tXx61//OmbOnBk1a9aMk046KbPt8ssvj1QqFUOHDo2XX345sz4/Pz/OPvvsWLduXZx44onRsWPHrP0NGDAgtt9++5gyZUpcddVVWduuuuqqmDJlSrRo0eJHDcgAAAAAAIDkSqXT6XRVNT5r1qz47LPPYuHChbF69erNJgD55z//GWeddVasXbs2mjVrFnvvvXfUrFkzPvroo/jmm28iJycn7rzzzjj//POz6t1yyy3xm9/8JlKpVBx88MGxzTbbxNtvvx3ffvttdOjQId55551o2rRpkf2NHj06jjjiiMjPz4/OnTtH586dY+LEiTFx4sSoV69ejBw5Mrp161ahY1i6dGk0atQolixZUuoQefxva3P5iOruArAZmHFD7+ruAgAAAADVrCK5QZWEQw888EDcfPPNMWnSpKz169aty1q+7rrr4s0334yWLVvG/fffX9ndKNX48ePj1ltvjbfeeiu++eabSKfTsd1228WBBx4YF110Ueyzzz7F1hs5cmTcfPPN8cEHH8SKFSuiVatWcdJJJ8WgQYOiQYMGJe5v6tSpce2118bIkSNj/vz50axZszj88MPjj3/8Y7Rr167C/RcOESEcAjYQDgEAAABQbeHQypUr46STTsoMu1a46VQqVSQceuWVV6JXr16RSqVi4sSJscsuu1RWV/7nCYeIEA4BGwiHAAAAAKhIblCpcw7169cvXnrppUin09G6desYNGhQkaHZCuvZs2c0a9YsIiJeeOGFyuwKAAAAAAAAxai0cOj111+PZ555JlKpVJx++ukxefLkuO666+KnP/1pyTvPyYmePXtGOp2Od955p7K6AgAAAAAAQAkqLRwaNmxYRETsuOOOMWzYsKhZs2a56u2xxx4REfH5559XVlcAAAAAAAAoQaWFQ6NHj45UKhX9+vUrdzAUEbH99ttHRMScOXMqqysAAAAAAACUoNLCoblz50ZERIcOHSpUr3bt2hER8f3331dWVwAAAAAAAChBpYVDubm5ERGxfv36CtVbuHBhREQ0bty4sroCAAAAAABACSotHGrevHlEREydOrVC9T7++OOIiGjZsmVldQUAAAAAAIASVFo4tP/++0c6nY5//etf5a6zYsWKeOqppyKVSsWBBx5YWV0BAAAAAACgBJUWDp188skRETFu3Lh44IEHylXnl7/8ZSxatCgiIn72s59VVlcAAAAAAAAoQaWFQ0cffXR069Yt0ul0nH/++TFkyJBYvnx5sWXHjRsXvXv3jkceeSRSqVT06tUr9tlnn8rqCgAAAAAAACVIpdPpdGU19vXXX8e+++4bc+bMiVQqFXl5edG8efP46quvIpVKRZcuXWLWrFkxb968iIhIp9PRqlWr+Oijj6Jp06aV1Y1EWLp0aTRq1CiWLFkSDRs2rO7uUE3aXD6iursAbAZm3NC7ursAAAAAQDWrSG5QaXcORUS0bNky3n///cwdRN9//33MnDkzUqlURESMHTs25s6dG+l0OtLpdOy7774xZswYwRAAAAAAAMCPpFLDoYgNAdGYMWPiueeeixNOOCG23nrrTBiUTqejfv360bt373jyySfj3Xffje23376yuwAAAAAAAEAJalRVw8ccc0wcc8wxERGRn58fixcvjvr16xsCDQAAAAAAoBpVWThUWN26daNu3bo/xq4AAAAAAAAoRaUPKwcAAAAAAMDmSzgEAAAAAACQIBUeVu6hhx6qin5ERES/fv2qrG0AAAAAAAA2IRwaMGBApFKpSu9IKpUSDgEAAAAAAFSxCodDERHpdLqy+wEAAAAAAMCPoMLh0NChQ6uiHwAAAAAAAPwIKhwO9e/fvyr6AQAAAAAAwI8gp7o7AAAAAAAAwI9HOAQAAAAAAJAgwiEAAAAAAIAEEQ4BAAAAAAAkSI2KVthxxx0jIiKVSsWXX35ZZP2m2rg9AAAAAAAAKl+Fw6EZM2ZExIYwZ+P1qVQq0un0JnVk4/YAAAAAAACofBUOh1q1alVskFPSegAAAAAAADYfm3znUHnXAwAAAAAAsPnIqe4OAAAAAAAA8OPZpHDo0EMPjUMPPTTGjBlT2f0BAAAAAACgCm1SODRq1Kh48803Y8GCBSWW6dKlS+y1117x5ptvbnLnAAAAAAAAqFwVnnOovD755JNIpVKxZMmSqtoFAAAAAAAAFWTOIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAgvygcCiVSlVKGQAAAAAAAH4cNX5I5eOOO67U7el0uswyBVKpVKxdu/aHdAcAAAAAAIAy/KBwKGJDAFScwncMlVQGAAAAAACAH9cmh0NlBT4CIQAAAAAAgM3PJoVD69evr+x+AAAAAAAA8CPIqe4OAAAAAAAA8OMRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEiTR4dDq1avjb3/7Wxx44IHRpEmTqF27drRo0SJ69eoVTzzxRLF1Ro4cGUcddVQ0bdo06tSpEx07dowrr7wyli9fXuq+pk6dGgMGDIgWLVpEXl5etGjRIgYMGBDTpk2rikMDAAAAAAAoVmLDoVmzZsWee+4ZF198cUyePDkOOOCAOO6446J169bx1ltvxVNPPVWkzi233BI9e/aMl19+OTp16hTHHHNMLFmyJK6//vro2rVrLFiwoNh9jR49OvbYY4948MEHo3HjxnH88cdH48aN48EHH4zdd9893nvvvao+XAAAAAAAgIiIqFHdHagOK1eujJ49e8akSZPimmuuiSuuuCJq1qyZ2Z6fnx9TpkzJqjNu3Li49NJLIzc3N55//vno1atXpmyfPn3i9ddfj/PPPz+efvrprHr5+flxyimnRH5+fgwaNCiuv/76zLYrrrgihgwZEqecckpMnjw56tSpU4VHDQAAAAAAkNA7h4YMGRKTJk2Kc889N66++uqsYCgiom7duvGTn/ykSJ10Oh1nnXVWJhgqKHv//fdHTk5OPPPMMzFp0qSsesOGDYvZs2dH+/btY/DgwVnbBg8eHO3bt4+vv/46Hnrooco9SAAAAAAAgGIkLhxas2ZN3H333RER8bvf/a5cdVavXh0jRoyIiIi+ffsW2d66des44IADIiJi+PDhWdsKlk877bTIycl+uXNycuLUU0+NiIhnn322AkcBAAAAAACwaRI3rNzYsWNjwYIFsf3228dOO+0Un376aTz77LMxe/bs2GqrraJ79+7Rq1evrCBnypQpkZ+fHxERXbt2Lbbdrl27xttvvx3jxo3LWl+wXFq9wuUAAAAAAACqUuLCoQkTJkRERIsWLeLyyy+Pv/zlL5FOpzPbb7zxxthzzz3jX//6V7Rq1SoiIqZPnx4REY0bN44GDRoU227Lli2zykZELFu2LL777ruIiExbJdWbP39+rFixIurVq1dsuVWrVsWqVasyy0uXLi37YAEAAAAAADaSuGHlCsKacePGxY033hgDBw6MyZMnx5IlS+K1116L9u3bx7hx46J3796xZs2aiNgQ8kREicFNRET9+vUjIju0KahXWt2CehvX3diQIUOiUaNGmUdBqAQAAAAAAFARiQuHCu4SWrNmTZx++ulxxx13RPv27aNhw4Zx+OGHx2uvvRa1a9eOiRMnxuOPP17Nvf0/gwYNiiVLlmQeX3/9dXV3CQAAAAAA2AIlLhwqPCzceeedV2R7q1atonfv3hERMXLkyKw6K1asKLHd5cuXR0REw4YNi91XSXUL6m1cd2N5eXnRsGHDrAcAAAAAAEBFJS4c2nHHHYt9XlyZb7/9NiIi2rRpExERixcvzhoqrrCCO3kKykZsCIeaNGkSEREzZ84stV7Tpk1LHbYOAAAAAACgMiQuHOrSpUukUqmIiFiwYEGxZQrWF8wH1KFDh6hbt25ERHz00UfF1ilY36VLlyL725R6AAAAAAAAVSFx4dC2224bBx54YET837Bxha1ZsybefPPNiIjYZ599IiKiVq1amaHmHn300SJ1vvrqqxgzZkxERBx//PFZ2wqWH3/88Vi/fn3WtvXr18cTTzwREREnnHDCJh8TAAAAAABAeSUuHIqIuPrqqyMiYsiQIfHee+9l1q9duzYuvfTSmDZtWjRo0CDOOuuszLbLL788UqlUDB06NF5++eXM+vz8/Dj77LNj3bp1ceKJJ0bHjh2z9jVgwIDYfvvtY8qUKXHVVVdlbbvqqqtiypQp0aJFi+jXr19VHCoAAAAAAECWVDqdTld3J6rD4MGD46qrrooaNWrEPvvsE9tuu22MHTs2ZsyYEXXq1Imnnnoqc7dQgVtuuSV+85vfRCqVioMPPji22WabePvtt+Pbb7+NDh06xDvvvBNNmzYtsq/Ro0fHEUccEfn5+dG5c+fo3LlzTJw4MSZOnBj16tWLkSNHRrdu3SrU/6VLl0ajRo1iyZIl0bBhwx/0WrDlanP5iOruArAZmHFD77ILAQAAAPA/rSK5QSLvHIqI+MMf/hCvvPJK9OzZMyZNmhTPP/98rFu3LgYMGBBjx44tEgxFRPz617+O1157LX7605/GhAkT4rnnnov69evHoEGD4sMPPyw2GIqIOOCAA2L8+PHRr1+/WLhwYTzzzDOxcOHC6NevX4wfP77CwRAAAAAAAMCmSuydQ1s6dw4R4c4hYAN3DgEAAADgziEAAAAAAACKJRwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ4BAAAAAAAkCDCIQAAAAAAgAQRDgEAAAAAACSIcAgAAAAAACBBhEMAAAAAAAAJIhwCAAAAAABIEOEQAAAAAABAggiHAAAAAAAAEkQ49P+77LLLIpVKRSqVisGDB5dYbuTIkXHUUUdF06ZNo06dOtGxY8e48sorY/ny5aW2P3Xq1BgwYEC0aNEi8vLyokWLFjFgwICYNm1aZR8KAAAAAABAiYRDETFmzJi4+eabI5VKlVrulltuiZ49e8bLL78cnTp1imOOOSaWLFkS119/fXTt2jUWLFhQbL3Ro0fHHnvsEQ8++GA0btw4jj/++GjcuHE8+OCDsfvuu8d7771XFYcFAAAAAABQROLDofz8/BgwYEBst912ceyxx5ZYbty4cXHppZdGbm5ujBgxIt5888148skn48svv4zDDjssJk+eHOeff36x7Z9yyimRn58fgwYNiokTJ8bjjz8eEydOjEGDBsWKFSvilFNOiZUrV1blYQIAAAAAAESEcCgGDRoUX3zxRdx7773RqFGjEssNGTIk0ul0nHXWWdGrV6/M+rp168b9998fOTk58cwzz8SkSZOy6g0bNixmz54d7du3LzJc3eDBg6N9+/bx9ddfx0MPPVS5BwYAAAAAAFCMRIdDo0aNittvvz3+v/buPEqq6twD9q8bZJRJUFBQIagE44iKUePFecaIc/y8iHLjkMQYjFHJdZ4SB6LRaIyJiBNXRZwxRk1wiNEoEVGcQAVBxQmQUVCgvj/8qE+kuwGFrm7redbqtarOec+u95TdpNK/3nv37ds3++yzT7V1n332WUaMGJEkOeKII5Y6v/7662eHHXZIktx9991LnFv8/PDDD09l5ZJvd2VlZQ477LAkyV133fX1bwQAAAAAAGA5lW04NHv27BxzzDFp3759rrjiihprx40bl7lz5yZJtt566yprFh8fPXr0EscXP1/R6wAAAAAAAFaFhqVuoFROOeWUTJgwIXfffXfatGlTY+2ECROSJK1bt06LFi2qrFl33XWXqE2SWbNmZerUqUmS9dZbr8brPvroo8yZMyfNmzevsm7+/PmZP39+8fnMmTNr7BkAAAAAAKAqZTlz6OGHH86f/vSnHH744TnggAOWWT9r1qwkqTa4SZLVV189yZKhzeLrarp28XVfvfarfvOb36RVq1bFr8WhEgAAAAAAwIoou3BoxowZ6d+/f9Zcc81cddVVpW5nuQ0cODAzZswofk2ePLnULQEAAAAAAPVQ2S0r94tf/CLvvPNObr/99rRr1265rlm8lNycOXOqrZk9e3aSpGXLlktdV9O1i6/76rVf1bhx4zRu3Hi5+gUAAAAAAKhO2YVDd999dxo2bJhrrrkm11xzzRLnXnvttSTJ9ddfn0cffTQdOnTIbbfdls6dOydJPvnkk8yaNavKfYcWz+RZXJt8EQ6tscYamTZtWiZNmpTNN9+82uvatWtX47J1AAAAAAAAK0PZhUNJsmDBgjz++OPVnp84cWImTpyY9ddfP0nSrVu3NGvWLHPnzs2oUaOy8847L3XNqFGjkiQ9evRY4niPHj3y6KOPZtSoUendu/dyXwcAAAAAALAqlN2eQ5988kkKhUKVX0cddVSS5Pzzz0+hUMjEiROTJI0aNcq+++6bJBk6dOhSY7799tv517/+lSTp06fPEucWP7/tttuyaNGiJc4tWrQot99+e5LkwAMPXHk3CQAAAAAAUI2yC4e+rtNPPz0VFRW54YYb8tBDDxWPz507N/3798/ChQtz0EEH5bvf/e4S1/Xr1y/rrLNOxo0blzPPPHOJc2eeeWbGjRuXTp06pW/fvrVyHwAAAAAAQHkry2Xlvo4ePXpk0KBBOfnkk7PPPvukV69eWWuttfLkk09mypQp6datW6699tqlrmvWrFnuuOOO7LHHHrnoooty3333ZZNNNsnYsWMzduzYNG/ePMOGDUvTpk1LcFcAAAAAAEC5MXNoBQwYMCCPPPJI9txzz7z44ou59957s/rqq2fgwIF57rnn0q5duyqv22GHHTJmzJj07ds306ZNy/DhwzNt2rT07ds3Y8aMyfe///1avhMAAAAAAKBcVRQKhUKpm2DFzZw5M61atcqMGTPSsmXLUrdDiXQ+fUSpWwDqgIm/3bfULQAAAABQYiuSG5g5BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEYalroBAADgm+l8+ohStwCU2MTf7lvqFgAAqEfMHAIAAAAAACgjwiEAAAAAAIAyIhwCAAAAAAAoI8IhAAAAAACAMiIcAgAAAAAAKCPCIQAAAAAAgDIiHAIAAAAAACgjwiEAAAAAAIAyIhwCAAAAAAAoI8IhAAAAAACAMiIcAgAAAAAAKCPCIQAAAAAAgDIiHAIAAAAAACgjwiEAAAAAAIAyIhwCAAAAAAAoI8IhAAAAAACAMlJ24dDnn3+ev//97/nVr36VbbbZJq1bt85qq62WDh06ZP/998+IESNqvP7RRx/NPvvsk3bt2qVp06b57ne/m//93//N7Nmza7zujTfeSL9+/dKpU6c0btw4nTp1Sr9+/fLWW2+tzNsDAAAAAACoUdmFQ48//nh22223XHbZZXnnnXfygx/8IAceeGDWXHPN3H///dlvv/1y3HHHpVAoLHXt5Zdfnt133z0PPfRQvve976V3796ZMWNGLrroomy99db5+OOPq3zNp556KptvvnluvPHGtG7dOn369Enr1q1z4403ZrPNNsszzzyzqm8bAAAAAAAgSRmGQ5WVlTnooIPyxBNPZMqUKXnggQdy++2356WXXsptt92WBg0a5LrrrsvNN9+8xHWjR4/OL3/5yzRo0CAjRozI448/njvuuCNvvvlmdt1117z++us5/vjjl3q9uXPn5tBDD83cuXMzcODAjB07NrfddlvGjh2bgQMHZs6cOTn00EPz6aef1tZbAAAAAAAAlLGyC4d22WWX3Hnnndlxxx2XOnfYYYelX79+SZKbbrppiXO/+c1vUigUcvTRR2fvvfcuHm/WrFmuv/76VFZWZvjw4XnttdeWuG7IkCF57733stFGG+WCCy5Y4twFF1yQjTbaKJMnT17q9QAAAAAAAFaFsguHlmXLLbdMkkyePLl47LPPPivuRXTEEUcsdc3666+fHXbYIUly9913L3Fu8fPDDz88lZVLvt2VlZU57LDDkiR33XXXSroDAAAAAACA6gmHvmL8+PFJkrXXXrt4bNy4cZk7d26SZOutt67yusXHR48evcTxxc9X9DoAAAAAAIBVoWGpG6hL3n///QwZMiRJctBBBxWPT5gwIUnSunXrtGjRospr11133SVqk2TWrFmZOnVqkmS99dar8bqPPvooc+bMSfPmzausmz9/fubPn198PnPmzOW5JQAAAAAAgCWYOfT/WbBgQY488sjMmDEjm266aY477rjiuVmzZiVJtcFNkqy++upJlgxtFl9X07WLr/vqtV/1m9/8Jq1atSp+LQ6VAAAAAAAAVoRw6P9z/PHH5+9//3vatm2bO++8M40aNSp1S0sYOHBgZsyYUfz68p5IAAAAAAAAy8uycklOOumkXH/99WnTpk0eeeSRbLTRRkucX7yU3Jw5c6odY/bs2UmSli1bLnVdTdcuvu6r135V48aN07hx4xruAgAAAAAAYNnKfubQL3/5y1x55ZVp3bp1Hn744Wy55ZZL1XTu3DlJ8sknnyyxVNyXLZ7Js7g2+SIcWmONNZIkkyZNqvG6du3a1bhsHQAAAAAAwMpQ1uHQqaeemt/97ndp1apVHn744Wy99dZV1nXr1i3NmjVLkowaNarKmsXHe/ToscTxxc9X9DoAAAAAAIBVoWzDodNPPz2XXnppWrVqlUceeSTbbLNNtbWNGjXKvvvumyQZOnToUufffvvt/Otf/0qS9OnTZ4lzi5/fdtttWbRo0RLnFi1alNtvvz1JcuCBB379mwEAAAAAAFhOZRkOnXHGGbn44ovTunXrZQZDi51++umpqKjIDTfckIceeqh4fO7cuenfv38WLlyYgw46KN/97neXuK5fv35ZZ511Mm7cuJx55plLnDvzzDMzbty4dOrUKX379l05NwcAAAAAAFCDhqVuoLbdd999ufDCC5MkG2ywQa6++uoq69q1a5fLLrus+LxHjx4ZNGhQTj755Oyzzz7p1atX1lprrTz55JOZMmVKunXrlmuvvXapcZo1a5Y77rgje+yxRy666KLcd9992WSTTTJ27NiMHTs2zZs3z7Bhw9K0adNVc8MAAAAAAABfUnbh0LRp04qPR40aVe1eQOuvv/4S4VCSDBgwIJtuumkGDRqUZ599NnPmzMl6662XgQMHZuDAgWnRokWVY+2www4ZM2ZMzj///Dz66KMZPnx41lxzzfTt2zdnnXVWunbtuvJuEAAAAAAAoAYVhUKhUOomWHEzZ85Mq1atMmPGjLRs2bLU7VAinU8fUeoWgDpg4m/3LXULQIn5TAD4PAAAwIrkBmW55xAAAAAAAEC5Eg4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAAAAAAlBHhEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUkYalbgAAAAAA+OY6nz6i1C0AJTbxt/uWugXqCTOHAAAAAAAAyohwCAAAAAAAoIwIhwAAAAAAAMqIcAgAAAAAAKCMCIcAAAAAAADKiHAIAAAAAACgjAiHAAAAAAAAyohwCAAAAAAAoIwIhwAAAAAAAMqIcAgAAAAAAKCMCIcAAAAAAADKiHAIAAAAAACgjAiHAAAAAAAAyohwCAAAAAAAoIwIhwAAAAAAAMqIcAgAAAAAAKCMCIcAAAAAAADKiHAIAAAAAACgjAiHAAAAAAAAyohwCAAAAAAAoIwIhwAAAAAAAMqIcAgAAAAAAKCMCIcAAAAAAADKiHAIAAAAAACgjAiHAAAAAAAAyohwCAAAAAAAoIwIh2rZsGHDstNOO6VNmzZp3rx5Nt9881xyySX5/PPPS90aAAAAAABQBoRDtegXv/hFDj300Dz11FPp2bNn9tprr0yaNCmnnXZadtlll3z66aelbhEAAAAAAPiWEw7VknvuuSe///3vs/rqq+ff//53/va3v2X48OEZP358Nt100/zzn//MmWeeWeo2AQAAAACAbznhUC256KKLkiSnn356evToUTzerl27XHPNNUmSP/zhD5kxY0ZJ+gMAAAAAAMqDcKgWvPvuu3nuueeSJEccccRS53/wgx9k3XXXzfz58/Pggw/WdnsAAAAAAEAZEQ7VgtGjRydJ1lhjjXTp0qXKmq233nqJWgAAAAAAgFWhYakbKAcTJkxIkqy33nrV1qy77rpL1H7V/PnzM3/+/OLzxcvPzZw5c2W1ST20aP7cUrcA1AH+twDwmQDweQBIfCYAfCYod4v/+xcKhWXWCodqwaxZs5IkzZs3r7Zm9dVXT1L9D+9vfvObnHvuuUsdXxwqAVC+Wl1R6g4AgFLzeQAASHwm4AuzZs1Kq1ataqwRDtUTAwcOzMknn1x8vmjRokybNi1t27ZNRUVFCTsDSmXmzJlZd911M3ny5LRs2bLU7QAAJeIzAQCQ+EwAfDFjaNasWVlnnXWWWSscqgUtWrRIksyZM6famtmzZydJtf9wN27cOI0bN17iWOvWrVdOg0C91rJlSx/6AACfCQCAJD4TQLlb1oyhxSpXcR8k6dy5c5Jk8uTJ1dYsPre4FgAAAAAAYFUQDtWCLbfcMkkyderUTJgwocqaUaNGJUl69OhRa30BAAAAAADlRzhUCzp16pRtttkmSTJ06NClzv/zn//M5MmT07hx4+yzzz613R5QTzVu3Dhnn332UktOAgDlxWcCACDxmQBYMRWFQqFQ6ibKwT333JM+ffpk9dVXz+OPP16cITR16tTsvPPOeemll/LLX/4yl112WYk7BQAAAAAAvs2EQ7XopJNOypVXXpnVVlstu+66a5o3b56///3v+eSTT7LDDjvkkUceSdOmTUvdJgAAAAAA8C0mHKpld9xxR66++uq88MIL+fzzz9O1a9cceeSRGTBgQBo1alTq9gAAAAAAgG854RAAAAAAAEAZqSx1AwAAAAAAANQe4RAAAAAAAEAZEQ4BAAAAAACUkYalbgAAAIAVM2PGjDz33HP56KOPsv7662f77bcvdUsAQAm89tpref311zNz5sxUt7V83759a7kroD6oKFT3rwYAAAB1yqxZszJgwIDcfPPNWbBgQZLkqKOOyuDBg5Mkf/nLX3LWWWfl7rvvzrbbblvKVgGAVeiZZ57Jsccem5dffrnamkKhkIqKiixcuLAWOwPqC8vKAQAA1AOffvppdtpppwwePDht2rTJ3nvvvdRfCO+333754IMPcs8995SmSQBglRs3blx23333jB07Nt///vfTpUuXJMnhhx+erbbaKg0aNEiS9OnTx6whoFrCIYB6aObMmbnmmmty5JFHZs8998wll1xSPDdu3Lg8/PDDmTdvXgk7BABWtt/97ncZPXp0fvSjH+XNN9/MAw88sFRNhw4d0r1794wcObIEHQIAteHiiy/OnDlzcs011+Spp57KjjvumCS59dZb8+yzz2b06NHZYostMn78+PzhD38ocbdAXSUcAqhnHn744XznO9/JiSeemKFDh+bRRx/Na6+9Vjz/+uuvZ++99859991Xwi4BgJXt9ttvT4cOHXL99denefPm1dZttNFGeeedd2qxMwCgNo0cOTJdu3bN8ccfX+X5733ve3nggQfy5ptv5sILL6zl7oD6QjgEUI+8+uqr6dOnT2bMmJETTjght99++1LLyey5555p1qxZ7r333hJ1CQCsCm+++WZ69uyZJk2a1FjXrFmzfPzxx7XUFQBQ26ZMmZJNNtmk+HzxMnKfffZZ8djaa6+dXr165a677qr1/oD6QTgEUI9cdNFFmTdvXm6//fb84Q9/yCGHHLJUTaNGjbLFFltkzJgxJegQAFhVGjRokM8//3yZde+8806NM4sAgPqtadOmadiwYfF5ixYtkiQffPDBEnUtW7bM5MmTa7U3oP4QDgHUIyNHjszmm2+eAw88sMa6Tp06ZcqUKbXUFQBQG7p27ZoxY8ZkwYIF1dbMnj07L774Yrp3716LnQEAtaljx46ZNGlS8fkGG2yQJHn66aeLxwqFQp5//vm0adOm1vsD6gfhEEA98tFHH2WjjTZaZt2CBQsyZ86cWugIAKgt+++/f6ZMmZILLrig2poLLrggM2bMSJ8+fWqxMwCgNm277bZ55ZVX8umnnyZJ9tprryTJgAEDMmLEiLz00ks54YQT8uabb2abbbYpZatAHSYcAqhHWrVqlXfffXeZdW+99VbWWmutWugIAKgtAwYMSMeOHXP++efngAMOyNChQ5N8sYTMXXfdlcMPPzyXXnppOnfuXO0G1QBA/bfPPvtk3rx5eeCBB5J8Mbv42GOPzZQpU7L//vtniy22yHXXXZdGjRrV+EclQHmrKHx1J3MA6qy99torTzzxRF577bWst956SZLKysr069cvgwcPTpKMHTs2W2yxRfr06ZNhw4aVsl0AYCV75ZVXsv/+++ett95KRUXFEucKhULWX3/9jBgxIhtvvHGJOgQASmHRokW54oorMmzYsEybNi3du3fPr3/96/Ts2bPUrQF1lHAIoB658847c+ihh2a77bbL8OHD06FDhyXCoY8//jj77bdfnnvuuTz44IPZc889S90yALCSzZs3L0OGDMmDDz6Yt956K4sWLcq6666bvffeO8cee2yaNWtW6hYBAIA6TjgEUM8cdthhGTZsWJo2bZoddtghjz76aDbccMN069Ytjz32WGbPnp3/5//5f3LzzTeXulUAAABgJXviiSfSoUOHZe5JPH78+EyZMiX/9V//VUudAfWJcAignlm4cGHOOuusXHHFFcXNJxdr1KhRTjzxxPz2t79NgwYNStQhAAAAsKpUVlbm6KOPzvXXX19j3Y9//OMMHjw4CxcurKXOgPqkYakbAGDFNGjQIBdeeGFOOeWUjBw5conlZHbdddestdZapW4RAAAAWIX8vT/wTQmHAOqpNm3a5MADDyx1GwBALfnOd76z3LUVFRV58803V2E3AEBdN3369DRp0qTUbQB1lHAI4FtkxowZadmyZSoqKkrdCgCwkk2cOHGZNRUVFSkUCj4LAMC3zKRJk5Z4Pnv27KWOLbZgwYK8/PLLefjhh9O1a9faaA+oh+w5BFCPjB07Nv/4xz+y1157LbHx5MiRI3PMMcdk0qRJWWONNXLppZemX79+pWsUAFjp3n777SqPL1q0KG+//XYeeOCBXHXVVRk4cGCOPvrorL/++rXcIQCwqlRWVhb/+GN5/xCkUCjkt7/9bU499dRV3R5QDwmHAOqRY489NoMHD86ECROy7rrrJkmmTp2a73znO5k1a1axrrKyMs8991y23HLLUrUKAJTAsGHD8qMf/SiPPPJIdt5551K3AwCsJJ07dy4GQpMmTUqzZs3Srl27KmsbNWqUTp065aCDDsoJJ5xgRjFQJeEQQD3yve99Lw0bNsyYMWOKx6644oqcfPLJOe6443LxxRfnvvvuS9++fdOvX78MHjy4hN0CAKWw6aabZu21187DDz9c6lYAgFWgsrLS/+cHvrHKUjcAwPL74IMPst566y1x7JFHHkmDBg1ywQUXpGXLljnyyCOz5ZZb5umnny5RlwBAKXXv3j3PPfdcqdsAAFaRG264If379y91G0A917DUDQCw/GbOnJlWrVotcezf//53tthii7Rt27Z4bMMNN8yDDz5Y2+0BAHXAu+++m88++6zUbQAAq8hRRx1V6haAbwEzhwDqkZYtW+bdd98tPn/11Vczbdq0bL/99kvVWlMYAMrPLbfckqeffjobb7xxqVsBAADqMDOHAOqRLbbYIk8++WTeeOONbLDBBrn++utTUVGRXr16LVE3YcKErL322iXqEgBYFY455phqz82aNSuvvfZaXnnllVRUVOSkk06qxc4AgNr2+eef58orr8ywYcPy+uuvZ+bMmVXWVVRUZMGCBbXcHVAfVBQKhUKpmwBg+QwbNiyHHXZYWrRoke985zt58cUXs9Zaa2XixIlp3Lhxki9+OdSuXbv07t07d955Z4k7BgBWlsrKZS/80LJly5x77rnCIQD4Fps/f3523XXXPP3001meX+0uWrSoFroC6hszhwDqkUMOOSSvvvpqLr744owZMyadO3fOTTfdVAyGkuSOO+7I559/vtRsIgCgfrvhhhuqPdeoUaN07NgxPXv2TJMmTWqxKwCgtv3+97/Pv/71r+y555658sorc+GFF+bmm2/OvHnzMn78+Nx888254oorcuqpp+bcc88tdbtAHWXmEEA99Nlnn2XmzJlp167dUucmTZqU6dOnp2vXrll99dVL0B0AAACwqmyzzTYZN25cJk2alFatWuXoo4/OTTfdlIULFxZr7r///hxwwAG5/fbbc/DBB5ewW6CuWva6BADUOY0aNaoyGEqS9dZbL5tvvrlgCAC+ZY455piceuqppW4DACixcePGZdttt02rVq2SfLGvUJIlwqHevXtnyy23zFVXXVWSHoG6TzgEAABQD9xyyy2ZMGFCqdsAAErs888/z5prrll83rRp0yTJzJkzl6jr1q1bXnrppVrtDag/7DkEUIedd955X/vaioqKnHnmmSuxGwCglDp06FD8y2AAoHx16NAhU6ZMKT5fe+21kySvvvpqtt9+++Lx9957b4nZRABfZs8hgDqssrIyFRUVWZF/qhfXV1RU+BAIAN8i/fv3z0MPPZSJEydmtdVWK3U7AECJ7LvvvnnppZcyadKkJMk//vGP7Lbbbtl///1z1113pbKyMo8//nh23XXX9OjRI88++2yJOwbqIjOHAOqws88+u9QtAAB1xDnnnJN77703P/7xj3PVVVelRYsWpW4JACiBPffcM3/961/z7LPPpmfPntlpp52y8cYb5/7770/Hjh2zzjrr5KWXXkqhUMhPfvKTUrcL1FFmDgEAANQD5513XsaNG5ehQ4embdu22W233dK5c+fiPgNfZnlZAPj2+vjjj/O3v/0tW2+9dbp165YkeeONN3LQQQcV9xhq0KBBfvazn+Xyyy8vZatAHSYcAgAAqIN22WWX7LXXXjn11FOTLN9ys5aXBYDy9vrrr2fatGnZaKON0rZt21K3A9RhlpUDAACogx577LF07ty5+Pyss85KRUVF6RoCAOq8xTOJAJZFOARQT7322mt5/fXXM3PmzGr/grhv37613BUAsKqcc845pW4BAAD4lhAOAdQzzzzzTI499ti8/PLL1dYsXk5GOAQAAAD120033fSNrve7AaAqwiGAemTcuHHZfffdM2fOnGy33Xb54IMPMmHChBx++OEZP358XnjhhSxcuDB9+vRJy5YtS90uAAAA8A3169fvay0t6w9HgZoIhwDqkYsvvjhz5szJNddck+OPPz5HH310JkyYkFtvvTVJ8vLLL6dv374ZP358nn766RJ3CwAAAHxTffv2XSocmj59eu67774kyWabbZYuXbokSSZOnJgXX3wxSbL//vunTZs2tdssUG9UFKrbqAKAOuc73/lOGjRokPHjxydJjj766Nx0001ZuHBhsWbKlCnZYIMNctJJJ+Wiiy4qVasAwDdUWVn5tf5KOEkqKiqyYMGCldwRAFAXTJ8+PT179ky7du1y7bXXZvPNN1/i/JgxY/KTn/wkH330UZ555pmsscYaJeoUqMsqS90AAMtvypQp2WSTTYrPGzRokCT57LPPisfWXnvt9OrVK3fddVet9wcArFyFQuFrfwEA305nnXVWpk2blr/+9a9LBUNJsvnmm+eBBx7Ixx9/nLPOOqsEHQL1gWXlAOqRpk2bpmHD//+f7hYtWiRJPvjgg6y77rrF4y1btszkyZNrvT8AYOXaa6+9ctppp5W6DQCgDrnvvvuy0047pXXr1tXWtGnTJjvvvHPuv//+/OEPf6i95oB6QzgEUI907NgxkyZNKj7fYIMNkiRPP/10MRwqFAp5/vnnrSsMAN8CHTp0SK9evUrdBgBQh3zwwQeprFz2glAVFRX58MMPa6EjoD6yrBxAPbLtttvmlVdeyaeffprki78mTpIBAwZkxIgReemll3LCCSfkzTffzDbbbFPKVgEAAIBVoEOHDhk5cmRmz55dbc2sWbPy2GOPZa211qrFzoD6RDgEUI/ss88+mTdvXh544IEkSdeuXXPsscdmypQp2X///bPFFlvkuuuuS6NGjXLBBReUuFsAAABgZTvggAMybdq09O7dO6+//vpS58eNG5cDDjgg06dPT58+fUrQIVAfVBTsVApQZ1155ZXZeOONs9tuu1Vbs2jRolxxxRUZNmxYpk2blu7du+fXv/51evbsWYudAgArW2VlZfr165fBgweXuhUAoA6ZPn16tt1227zxxhuprKxMjx490qVLlyTJxIkT8/zzz2fhwoXZYIMN8u9//9uy80CVhEMAdVhNvxTaZZddstdee+XUU08tQWcAwKomHAIAqvPhhx/mpz/9ae6+++4sWrRoiXMVFRXp06dPrr766rRv375EHQJ1XcNSNwDA1/PYY4+lc+fOpW4DAFhFvvqLHgCAxdZaa60MGzYs77zzTp544om88847SZKOHTvmv/7rv7LuuuuWuEOgrhMOAQAAAADUQ506dcoRRxxR6jaAeqiy1A0AAAAAAABQe4RDAAAAAAAAZUQ4BAAAAAAAUEbsOQRQx73xxhu56aabVvhckvTt23dVtQUAAAAA1FMVhUKhUOomAKhaZWVlKioqvta1FRUVWbBgwUruCAAAAACo78wcAqjD1ltvva8dDgEAAAAAVMXMIQAAAAAAgDJSWeoGAAAAAAAAqD2WlQMAAAAAqGemTp2aP//5zxk5cmTefffdJEnHjh2zyy675H/+53/Stm3bEncI1GWWlQMAAAAAqEcefvjh/OhHP8onn3ySr/56t6KiIm3atMnQoUOzxx57lKhDoK4TDgEAAAAA1BPjx4/Plltumblz52azzTbL0Ucfna5duyZJ3nrrrQwZMiQvvPBCmjVrltGjR2fDDTcsccdAXSQcAgAAAACoJ/r3758bbrgh55xzTs4666wqa84///ycffbZOeaYY/KXv/ylljsE6gPhEAAAAABAPdG5c+c0bdo0r776ao113bt3z6effpqJEyfWTmNAvVJZ6gYAAAAAAFg+77//fnr06LHMuh49euT999+vhY6A+kg4BAAAAABQTzRv3jwffvjhMus+/PDDNG/evBY6Auoj4RAAAAAAQD2xxRZb5IknnshLL71Ubc2LL76Yxx9/PFtssUXtNQbUK8IhAAAAAIB64sc//nE+//zz7Lbbbrnmmmsye/bs4rnZs2fnD3/4Q3bfffcsXLgwxx57bAk7BeqyikKhUCh1EwAAAAAALJ+jjjoqN998cyoqKpIkbdu2TZJMnTo1SVIoFNK3b98MGTKkVC0CdZxwCAAAAACgnrn22mtz2WWX5a233lrieNeuXXPKKafkuOOOK1FnQH0gHAIAAAAAqKfefffdvPvuu0mSjh07pmPHjiXuCKgPhEMAAAAAAABlpLLUDQAAAAAAAFB7Gpa6AQAAAAAAqnbTTTd9o+v79u27kjoBvk0sKwcAAAAAUEdVVlamoqLia1+/cOHCldgN8G1h5hAAAAAAQB21yy67rHA49PTTT2fu3LnfKFQCvt2EQwAAAAAAddSjjz663LVPPvlkTj311Hz66adJkk033XRVtQXUc5WlbgAAAAAAgK9v7Nix6d27d3baaaf8+9//zrrrrpshQ4Zk9OjRpW4NqKPMHAIAAAAAqIcmT56cM888M7feemsWLlyYtm3b5te//nV++tOfplGjRqVuD6jDhEMAAAAAAPXI9OnTc+GFF+aaa67JvHnz0qxZs5x00kk57bTT0rJly1K3B9QDFYVCoVDqJgAAAAAAqNm8efNy+eWX55JLLsnMmTPToEGDHHPMMTnnnHPSoUOHUrcH1CNmDgEAAAAA1GGLFi3KX/7yl5x33nmZMmVKCoVCDjzwwFx00UXZaKONSt0eUA8JhwAAAAAA6qi77ror//u//5tx48alUCikV69eufjii9OzZ89StwbUY5aVAwAAAACooyorK1NRUVHcV2ifffZZoeu33377VdQZUJ8JhwAAAAAA6qjF4dDXUVFRkQULFqzkjoBvA8vKAQAAAADUUeutt97XDocAqmPmEAAAAAAAQBmpLHUDAAAAAAAA1B7hEAAAAAAAQBkRDgEAAAAAAJQR4RAAAAAAAEAZEQ4BAAAAAACUEeEQAAAAAABAGREOAQAArEJ33HFH9tprr7Rv3z6rrbZaKioqUlFRkSFDhiRJhgwZstSxVWXixInF1+rXr983Hm+nnXYqjlcXxvm26Ny5cyoqKtK5c+dStwIAwLeUcAgAAFbQ4l9ir+gvsr/udXVZv379vlawsazrbrrppuL5jTfeOPPnz1/mmIVCIXvssUfxurPPPnupmi+HEI899thy9/t1/exnP8thhx2Wv/3tb/nwww+zYMGCVf6asKp98sknuf7663PQQQelS5cuad68eZo0aZIuXbrkv//7v/Ovf/2r1C0CALAMDUvdAAAAwFf17ds3w4cPz3333ZdXX301Z555Zi655JIar/njH/+YRx55JEmy5ZZb5owzzqiNVqs1evToXH311UmStm3b5sQTT8x3v/vdNG7cOEnSo0ePUrYHX8vll1+es88+O7NmzVrq3MSJEzNx4sTceuutOeWUU5b5MwsAQOkIhwAAgDrpuuuuy1NPPZWpU6dm0KBBOeCAA7L99ttXWfvmm2/m1FNPTZI0atQoN910U1ZbbbXabHcpI0aMKD6+8sorc8QRR1RZ169fv5WyxBvUhnvvvTezZs1KkyZNss8++6RXr15p37593nvvvdx999158sknUygUcumll6ZLly454YQTSt0yAABVsKwcAABQJ7Vv374482bRokXp169f5s6du1Td4nNz5sxJkpx77rnZZJNNarXXqkyePLn42Cwhvi2aNGmS008/PZMmTcrw4cPz85//PIcddlgGDBiQJ554Iqeddlqx1swhAIC6SzgEAADUWYcddlgOOeSQJMn48eMzcODApWoGDRqUf/7zn0mS73//+/nVr35Vqz1W58v7JDVp0qSEncDKc/vtt+c3v/lN1lxzzSrPn3POOcVZexMnTsy0adNqsz0AAJaTcAgAAOqQefPm5d57783Pf/7zbL/99llzzTWz2mqrpUWLFtlwww3z3//938V9dZZl5syZGTRoUHbeeee0b98+jRo1SosWLdK5c+dss8026d+/f4YNG5bPPvtsFd/VN3PNNddkrbXWSpJcddVVefzxx4vnXnnllZx55plJkqZNm+bGG29MgwYNStJnkjz22GOpqKhIRUVFbrzxxuLxLl26FI9XVFQssYzckCFDiseHDBmyzNe4//7707dv32ywwQZp0aJFmjVrli5duuTII4/Mo48+utLuZfr06TnzzDOz6aabZvXVV0/r1q2zxRZb5LzzzsvUqVNX2utU5+67786+++6bjh07pnHjxllnnXVy0EEH5YknnljuMWbMmJFBgwZlt912yzrrrJPGjRtnjTXWyFZbbZWBAwfm3XffXeYY77zzTq655pocfvjh2XjjjdOiRYusttpqadeuXbbddtsMHDhwiVliyzPegAED0q1btzRr1qw4zqBBg/Lpp58u9ziLvfnmmzn99NOzzTbbZM0110yjRo3Svn377LLLLvn9739f5Wy7L+vcuXMqKirSuXPnJF+EmldffXV22mmnrL322mnQoEHxXJK0atWqxvGaNGmS1Vdfvfi8stKvHQAA6qQCAACwQpIUv1b2dV26dFmirrqvH/7wh4VZs2ZVO86oUaMKHTp0WK6xnnvuuRW6jy876qijiuPccMMNq+y6u+++u1jfpUuXwqxZswqff/55Yauttioev/zyy5c5Tq9evYr1I0eOXO5+l9fIkSOX6z0/6qijitfccMMNy/VeTJo0qbDddtstc+yDDjqoMGfOnCrHmDBhQpU9fNWzzz5baN++fbWvse666xbGjBmzxPv5TXx5nE8//bRw8MEH13iPl1xyyTLHvOOOOwprrLFGjeM0adKkMGTIkGrHGDlyZKGiomKZ73mjRo0Kf/nLX5bZ0wMPPFBo0aJFteNssskmhUmTJhXWX3/9QpLC+uuvX+1YCxcuLAwcOLDQsGHDGnvr1KlTYdSoUdWO8+XXmjBhQmGTTTZZaoya+qjqHhdft9lmmy33dQAA1K6GVUdGAABAKcydOzetW7fOLrvski233DLrr79+mjVrlpkzZ+bFF1/M7bffnilTpuTee+/NMccckzvuuKPKMQ444IC8//77SZKtttoqffr0SceOHdO8efNMnz49r776akaOHJkxY8bU9i1+LQcccECOPPLI3HLLLZkwYUJ+9atfpUOHDvnPf/6TJOnVq1dOOumkEneZbLLJJrn77ruTJFdeeWVGjhyZJPnTn/5UnP2UJOutt94KjTt58uRsu+22mTJlSpJkyy23zAEHHJANNtgglZWVef3113PTTTflrbfeyvDhwzNnzpw8+OCDqaioWOF7mDBhQvbYY4988sknSZJu3bqlX79+6dKlSz7++OPcdddd+cc//pEDDjhgmbNIvo7+/fvnzjvvzCabbJIf/ehH6dq1a+bMmZP7778/99xzT5LktNNOy3bbbZcf/OAHVY7x5z//Occdd1wKhUIaNWqUH/7wh/mv//qvtG/fPrNnz84///nPDB06NPPmzUu/fv3SqFGj/OhHP1pqnHnz5qVQKKRbt27Zeeeds/HGG6ddu3Zp2LBh3n///TzxxBO555578tlnn+XHP/5x2rdvn/3226/Knp555pkceOCBxZl6W2+9dY444oiss846ee+99zJ06NCMGjUqhxxySD7//PNlvk9HHXVUbrnlliTJGmuskcMOOyxbbbVVWrZsmQ8//DAjRozIX//617zzzjvZeeedM2rUqGy00UbVjjd//vwceOCBGTt2bL7//e/n4IMPTqdOnTJt2rS8/PLLy+wnSV577bX079+/+Pyiiy5arusAACiBUqdTAABQ32QVzhx68MEHC5999lm15+fMmVM44IADiuM8+eSTS9UMGzaseP7kk0+usaeXX3658OGHHy7/TXxFbc0cKhQKhWnTphXWWWed4nWLZ0ysvvrqhbfeemu5xljVM4e+7Mv3OGHChGrrljVzaNGiRcUZQw0aNChcd911VY4zb968wuGHH14c689//vNSNcszc2iPPfYo1hx66KGF+fPnL1Xzu9/9bqnZJd/El/+7LP6+Xbhw4VJ1559/frGmd+/eVY41ZsyYQqNGjQpJChtuuGHh1VdfrbLulVdeKX4/tWjRojB16tSlaiZOnFh44YUXaux99OjRhbXWWqv4eosWLVqqZsGCBYXu3bsXex8wYMBS97dw4cLCgAEDlmvGzrXXXrvE+zB9+vQq64YPH178Odlhhx2qrFk8c2jx129/+9sa77c6Y8aMKb4PSQrnn3/+1xoHAIDaYfFfAAD4Br68j8yyvpbH3nvvXdzMvSrNmjXLjTfemObNmyfJEvvaLPbGG28UH3/5r/irsvHGG1e7sfyKOvroo5f7vaiq72Vp06ZNrrvuuuLzBQsWJEkuvfTSdOnSZaXcQ110//335+mnn06SnHPOOfnxj39cZV3jxo1z4403FveHGTRo0Aq/1osvvpiHH344yRezm4YMGZJGjRotVTdgwIAcfPDBKzz+8ujVq1cuu+yyKveqGThwYDp27JgkeeSRR4rfA192zjnn5LPPPkuTJk3y4IMP5rvf/W6Vr9O9e/fiHk+zZs3Kn//856Vq1l9//Wy++eY19rvFFlsUZ8iMHz8+//rXv5aqGTFiRF599dUkSc+ePTNo0KCl7q+ysjKDBg1Kz549a3y9+fPn59xzzy3ew5133pnWrVtXWXvggQfm1FNPTZI89dRT+fe//13j2D/84Q9z2mmn1VhTlTFjxmTnnXfOhx9+mOSL/wZnnHHGCo8DAEDtEQ4BAEA907Jly2y66aZJvliq6qsWB0dJisuufVv84Ac/SJs2bYrPN9xwwxx//PEl7GjVWxykNW7cOD//+c9rrP3y8mivvfZaJk2atEKvdddddxUfn3DCCWnatGm1tb/61a9WaOzlNWDAgGrD1AYNGmTnnXdO8sWSb2+++eYS5z/55JPce++9SZI+ffpkgw02qPG1dt9996y99tpJkr/97W9fu+cvL29X1c/kl9/Xmu6voqIiv/zlL2t8rYcffri4vOAvfvGLKsO7LzvqqKOKj5d1j8v6/qrKlClTsvvuu2fatGlJvgglzz777BUeBwCA2mXPIQAA+AYW7y+zPPr06bNcddOnT8+tt96ahx56KGPHjs3UqVMzZ86cFAqFpWrfeeedpY7ttttuqaioSKFQyAknnJA33ngjP/rRj6qdQbGynHjiidlll12Wq/bL+/GsiJNOOinTp08vPh8/fnzuvffe/PCHP1zhseqLJ554IknSvn37/OMf/1hm/Zffn1deeWWF9jd69tlni4933XXXGmu32WabtGjRIrNmzVru8ZfHdtttV+P5Tp06FR9/+V6TL2bHLFq0KMkXYdriPYpq0qJFi0yZMiWvvPJKtTUvvPBCbrnlljz99NMZP358Zs6cmfnz51dZW9XP5Iq8r8s6v/j7IflixtOy7vHL+xfVdI8NGjTI9ttvX+NYVTnzzDPz0UcfJUlOP/30nHzyySs8BgAAtU84BAAA38ABBxywUse79957079//0ydOnW56mfOnLnUse7du+eMM87I+eefnzlz5uS8887Leeedl7XXXjvbb799dtxxx+y1117p1q3bSu29R48ey/1+LM8v7b/q/vvvL86iWXPNNYu/kD7++OOz4447Zo011ljhMeu6OXPm5OOPP06STJo0abkDxsUWz+ZYXu+9917x8bJm3VRUVKRr16554YUXVug1lqVdu3Y1nm/cuHHx8bx585Y4N3HixOLjIUOGFJeNWx5VvVcLFizIT3/60/z5z3+uMpytSlU/k4vf15YtWy5zGce2bdumdevW+eSTT6o8/+V7POWUU5arp8Vq+n5o27ZtmjRpskLjJf//bKSGDRtaSg4AoB4RDgEAQB3x9NNP5+CDDy7uo7LZZptlt912ywYbbJA2bdqkcePGxeWozjjjjLz88svFWRJfdd5552WbbbbJb3/72+IeKFOmTMnw4cMzfPjwJMkOO+yQQYMGZdttt62Fu/tmpk2blmOPPTbJF6HE7bffnttuuy3XXXdd3n///Zx44om59dZbS9zlylddQLC8PvvssxWqnz17dvFxs2bNlln/5SUMV5aq9hpaXt/k/fryDJvFTjrppOI+V6uttlr22muv9OzZM506dUrz5s2L+4N9+OGHOe6445IkCxcuXGqcxe/r8rynyRfva3X38k3usabvh5qWEKzJ+++/n+SLUG9VfD8AALBqCIcAAKCOOOuss4rB0NVXX52f/OQn1dZeeOGFyxyvd+/e6d27dz744IM8+eSTefrpp/P444/n+eefT6FQyFNPPZUdd9wxDz74YHbbbbeVdh+rwk9/+tPiL6F/+tOfZuedd87WW2+dhx56KJMmTcrQoUNz6KGHfuuWl1t99dWLj3v06LHK95D68uvNnTt3iVk6VZkzZ84q7WdFfbn/K6+8MieeeOLXHmvy5Mm59tprkyQdO3bMyJEjs+GGG1ZZ+/LLLy+zrxkzZmTu3LnL9do1va9fvscXX3yxuP9YqSz+N6tBgwYl7QMAgBXz9f8kCwAAWGk+//zzPPbYY0mSrbbaqsZgKFlyaallad++fQ4++OAMGjQoo0aNysSJE3PIIYcUX3fAgAFft+1aceedd+a2225LknTt2jUXX3xxki/2irn++uuLdccff/wKL6NW17Vq1aoYBlS1l83K1rFjx+LjN954o8baQqGQt956a1W3tEK+vB/R5MmTv9FYjz76aHFm3umnn15tMJQkEyZMqHGsxe/rzJkzi8shVmfq1Kk1zg5amfe4MrRv3z7t27fPWmutVepWAABYAcIhAACoAz7++OPiX+Ava6+X5557rrgPzdex3nrr5dZbby3ufTJ27NhvvHzZqvLhhx/mhBNOSPLFcmM33HDDEktz7bbbbsXlvBYvL/dt06tXryRfvBereuZQz549i4//8Y9/1Fj73HPPVbm/TintuOOOxaUXH3rooW801uKZasmyfyb/+te/1nh+Rd7Xv//97zWeX/z9sDyvWxvef//9vP/++3n++edL3QoAACtAOAQAAHXAl/fqWNaMjbPPPvsbv95qq622xCyRxcFUXXP88ccXg7CTTjopO+6441I1l156adZff/0kydChQ3PvvffWao+r2lFHHVV8fMYZZ6RQKKyy1+rTp0/x8R//+MfMmzev2tpBgwatsj6+rrXWWit77713kuSll17K//3f/33tsZb3Z/Ktt97KjTfeWONYX35fr7jiimr/GxYKhVx++eU1jrX33nsXg93Bgwcv898LAACoinAIAADqgJYtW2ajjTZKkvznP//JnXfeuVTNwoULM2DAgGXOFrjyyiszbNiwGjeff+qpp/Liiy8m+WKZqnbt2n2D7leNW2+9NXfffXeSZKONNqp2n6XFy8stnjHybVte7uCDD862226b5IvZMH379s3s2bOrrV+4cGEeeuihXHDBBSv8Wptttln22GOPJMnbb7+dY445Jp9//vlSdVdddVXuuOOOFR6/Nlx44YVp1KhRkuR//ud/lhkQTZs2Lb/73e/y6KOPLnF8m222KT6+7LLLMnXq1KWunTRpUnr37r3MvZf23XffdO/ePUnyzDPP5NRTT10qICoUCjn11FPzzDPP1DhW8+bNc8455yT5Yl+oPffcM6NHj67xmjfeeCMnn3xyPvzwwxrrVtTEiRNTUVGRioqKJZa7AwCg7mtY6gYAAIAv/OIXvyjuNXTooYfmsMMOS69evdKmTZu88cYbufXWW/Pqq69mk002SePGjatdYuz555/PjTfemFatWmXPPfdMjx490rFjxzRq1CgffPBBHn/88dx3333F/VR+/etf19o9Lq8pU6YUl4irrKzMkCFD0rRp02rrd9111xx33HG59tpri8vL3XrrrTW+xvXXX79UIFCdM844I02aNFn+G1iJKioqMnz48Gy33XaZPHlybrnllowYMSKHHHJIttpqq6yxxhqZN29e3nvvvYwZMyaPPPJIPvroo+y6664544wzVvj1rr322vTo0SOffPJJ/u///i+jR49Ov3790qVLl0ydOjXDhw/P3//+93Tp0iWtWrXKCy+8sPJv+hvYYost8qc//Sn9+/fP3Llzc8QRR+SSSy5J7969s+GGG6Zp06aZMWNG3njjjTz77LN54oknsmDBgtx8881LjLPddttl2223zb///e+8/fbb+e53v5tjjz023bt3z8KFC/PMM8/k5ptvzpw5c9KvX78MGTKk2p4aNGiQwYMHp1evXvnss89y2WWX5fHHH88RRxyRddZZJ++9916GDh2a5557Lttuu20mT56c9957r9rxfvKTn+Q///lPBg8enLfeeitbbbVV9txzz+y6667p1KlTKioqMm3atLz66qt58skni/+NTj755JXxFgMA8C0gHAIAgDri+OOPz6hRozJ48OAUCoXcdtttue2225ao2XTTTXPvvffm6KOPrnacxTNoZsyYkTvuuKPaGR6rrbZazj777OKePnXJj3/840yfPj1J8stf/jLbbbfdMq+59NJL89BDD2XixIkZOnRoDj300Pzwhz+stv6WW25Z7n5OOeWUkoVDSdKxY8eMGjUq/fr1y1//+tdMnz491113XY3XfN2ZHF26dMnf/va39O7dOx9++GFee+21nH766UvUrLvuurnnnnvy85///Gu9xqrWr1+/rL322jnmmGPy3nvv5YUXXqgxxGrcuHGVs+duu+227LLLLpkwYUI+/vjjXHTRRUvVnHjiiRkwYECN4VCSfP/738/w4cNzxBFHZNasWXnuuefy3HPPLVGz6aabZtiwYVUun/hVf/nLX9KtW7ece+65mTt3bh566KEa91lq165dSb+HAQCoW4RDAABQR1RUVOT666/Pvvvum+uuuy6jRo3KzJkz07Zt23Tr1i2HHHJI+vfvv8xf8P7xj3/M4YcfnpEjR2bUqFEZN25cPvrooyxYsCAtW7bMhhtumJ122in9+/fPhhtuWEt3t/wGDx6cESNGJEm6d++e888/f7muW3311XP99ddnt912S6FQyPHHH58dd9wxa6yxxqpst9astdZaefDBB/PMM8/k1ltvzT//+c9Mnjw5n3zySZo0aZIOHTqke/fu+cEPfpD99tsv3/ve9772a/Xs2TOvvfZaBg0alHvuuScTJkxIw4YN07lz5xx44IH52c9+lrZt267Eu1v59txzz7z11lsZOnRoHnzwwfznP//JRx99lHnz5qVFixbp3LlzNt988+yyyy7p3bt3WrduvdQYnTt3zujRo3PFFVfkrrvuKu7v06FDh2y//fbp379/dtppp0ycOHG5etpvv/3yyiuv5LLLLsuDDz6YyZMnp1mzZunatWsOO+ywnHDCCWnWrNlyjVVRUZFTTz01Rx99dAYPHpxHH300r7zySnH5u9atW2eDDTbI1ltvnd133z177LFHVlttteUaGwCAb7+KwqrczRQAAAAAAIA6pbLUDQAAAAAAAFB7hEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABlRDgEAAAAAABQRoRDAAAAAAAAZUQ4BAAAAAAAUEaEQwAAAAAAAGVEOAQAAAAAAFBGhEMAAAAAAABl5P8F4toUGUdZbP8AAAAASUVORK5CYII=",
      "text/plain": [
       "<Figure size 2000x1000 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "resources_data_details = pd.read_pickle(f\"./data/{country}_data_details.pkl\")\n",
    "plot_sorted_bar_chart(\n",
    "    resources_data_details,\n",
    "    \"has_hxl_header\",\n",
    "    \"counts\",\n",
    "    f\"{country} Datafiles with HDX Headers\",\n",
    "    \"Has HXL field header?\",\n",
    "    \"Files\",\n",
    "    rotation=90,\n",
    "    figsize=(20, 10),\n",
    ")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So we can see a lot of HXL tags seem to be missing for this country."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Predicting HXL field tags\n",
    "\n",
    "OK, so we need to predict some tags. \n",
    "\n",
    "### Downloading data with HXL tags\n",
    "\n",
    "First we will download data likely to have HXL tags by searching for 'HXL'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets_hxl = pd.DataFrame(Dataset.search_in_hdx(\"HXL\"))\n",
    "print(datasets_hxl.shape)\n",
    "display(datasets_hxl.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "datasets_hxl.to_excel(f\"./data/datasets_hxl.xlsx\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, we download ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "output_folder = f\"./data/hxl_datasets\"\n",
    "if not os.path.exists(output_folder):\n",
    "    os.mkdir(output_folder)\n",
    "\n",
    "download_data(datasets_hxl, output_folder, agent_count, errors=\"ignore\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Extract Meta data for downloaded data that has HXL tags"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "hxl_resources_data_details = extract_all_data_details(f\"./data/hxl_datasets/\", [])\n",
    "hxl_resources_data_details.to_pickle(f\"./data/hxl_data_details.pkl\")\n",
    "print(\"Done\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>resource_id</th>\n",
       "      <th>resource_name</th>\n",
       "      <th>resource_format</th>\n",
       "      <th>dataset_id</th>\n",
       "      <th>dataset_name</th>\n",
       "      <th>dataset_org_title</th>\n",
       "      <th>dataset_last_modified</th>\n",
       "      <th>dataset_tags</th>\n",
       "      <th>dataset_groups</th>\n",
       "      <th>dataset_total_res_downloads</th>\n",
       "      <th>...</th>\n",
       "      <th>type</th>\n",
       "      <th>dataset</th>\n",
       "      <th>sheet</th>\n",
       "      <th>resource</th>\n",
       "      <th>num_rows</th>\n",
       "      <th>num_cols</th>\n",
       "      <th>columns</th>\n",
       "      <th>first_row</th>\n",
       "      <th>has_hxl_header</th>\n",
       "      <th>first_nrows</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>bd15032b-57b2-47ca-87e0-09a6c1a58d1f</td>\n",
       "      <td>Literacy Data for Rwanda</td>\n",
       "      <td>CSV</td>\n",
       "      <td>fa20a295-e75b-4ac2-b00e-65bb43626faa</td>\n",
       "      <td>dhs-data-for-rwanda</td>\n",
       "      <td>The DHS Program</td>\n",
       "      <td>2020-03-20T23:05:25.396054</td>\n",
       "      <td>[{'display_name': 'demographics', 'id': '7aa60...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Rwanda',...</td>\n",
       "      <td>5480</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '6e71bebe-0a04-4c...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>108</td>\n",
       "      <td>25</td>\n",
       "      <td>['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...</td>\n",
       "      <td>['#country+code', '#meta+id', '#indicator+name...</td>\n",
       "      <td>True</td>\n",
       "      <td>ISO3    DataId                    ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>a4ce98da-69b5-499b-8f9d-fd1b7e6685f0</td>\n",
       "      <td>Anemia Data for Rwanda</td>\n",
       "      <td>CSV</td>\n",
       "      <td>fa20a295-e75b-4ac2-b00e-65bb43626faa</td>\n",
       "      <td>dhs-data-for-rwanda</td>\n",
       "      <td>The DHS Program</td>\n",
       "      <td>2020-03-20T23:05:25.396054</td>\n",
       "      <td>[{'display_name': 'demographics', 'id': '7aa60...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Rwanda',...</td>\n",
       "      <td>5480</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '6e71bebe-0a04-4c...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>61</td>\n",
       "      <td>25</td>\n",
       "      <td>['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...</td>\n",
       "      <td>['#country+code', '#meta+id', '#indicator+name...</td>\n",
       "      <td>True</td>\n",
       "      <td>ISO3    DataId                    ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>74d7b898-dacb-4538-a7e5-51797f209e10</td>\n",
       "      <td>HIV Prevalence Data for Rwanda</td>\n",
       "      <td>CSV</td>\n",
       "      <td>fa20a295-e75b-4ac2-b00e-65bb43626faa</td>\n",
       "      <td>dhs-data-for-rwanda</td>\n",
       "      <td>The DHS Program</td>\n",
       "      <td>2020-03-20T23:05:25.396054</td>\n",
       "      <td>[{'display_name': 'demographics', 'id': '7aa60...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Rwanda',...</td>\n",
       "      <td>5480</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '6e71bebe-0a04-4c...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>94</td>\n",
       "      <td>27</td>\n",
       "      <td>['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...</td>\n",
       "      <td>['#country+code', '#meta+id', '#indicator+name...</td>\n",
       "      <td>True</td>\n",
       "      <td>ISO3    DataId                    ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9e384243-f0c3-461d-9eff-2e2cb335252d</td>\n",
       "      <td>MDGs Data for Rwanda</td>\n",
       "      <td>CSV</td>\n",
       "      <td>fa20a295-e75b-4ac2-b00e-65bb43626faa</td>\n",
       "      <td>dhs-data-for-rwanda</td>\n",
       "      <td>The DHS Program</td>\n",
       "      <td>2020-03-20T23:05:25.396054</td>\n",
       "      <td>[{'display_name': 'demographics', 'id': '7aa60...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Rwanda',...</td>\n",
       "      <td>5480</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '6e71bebe-0a04-4c...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>179</td>\n",
       "      <td>27</td>\n",
       "      <td>['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...</td>\n",
       "      <td>['#country+code', '#meta+id', '#indicator+name...</td>\n",
       "      <td>True</td>\n",
       "      <td>ISO3    DataId                    ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>90992dc5-37d8-453e-ad34-6963eb475efa</td>\n",
       "      <td>SDGs Data for Rwanda</td>\n",
       "      <td>CSV</td>\n",
       "      <td>fa20a295-e75b-4ac2-b00e-65bb43626faa</td>\n",
       "      <td>dhs-data-for-rwanda</td>\n",
       "      <td>The DHS Program</td>\n",
       "      <td>2020-03-20T23:05:25.396054</td>\n",
       "      <td>[{'display_name': 'demographics', 'id': '7aa60...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Rwanda',...</td>\n",
       "      <td>5480</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '6e71bebe-0a04-4c...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>118</td>\n",
       "      <td>27</td>\n",
       "      <td>['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...</td>\n",
       "      <td>['#country+code', '#meta+id', '#indicator+name...</td>\n",
       "      <td>True</td>\n",
       "      <td>ISO3    DataId                    ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                            resource_id                   resource_name  \\\n",
       "0  bd15032b-57b2-47ca-87e0-09a6c1a58d1f        Literacy Data for Rwanda   \n",
       "1  a4ce98da-69b5-499b-8f9d-fd1b7e6685f0          Anemia Data for Rwanda   \n",
       "2  74d7b898-dacb-4538-a7e5-51797f209e10  HIV Prevalence Data for Rwanda   \n",
       "3  9e384243-f0c3-461d-9eff-2e2cb335252d            MDGs Data for Rwanda   \n",
       "4  90992dc5-37d8-453e-ad34-6963eb475efa            SDGs Data for Rwanda   \n",
       "\n",
       "  resource_format                            dataset_id         dataset_name  \\\n",
       "0             CSV  fa20a295-e75b-4ac2-b00e-65bb43626faa  dhs-data-for-rwanda   \n",
       "1             CSV  fa20a295-e75b-4ac2-b00e-65bb43626faa  dhs-data-for-rwanda   \n",
       "2             CSV  fa20a295-e75b-4ac2-b00e-65bb43626faa  dhs-data-for-rwanda   \n",
       "3             CSV  fa20a295-e75b-4ac2-b00e-65bb43626faa  dhs-data-for-rwanda   \n",
       "4             CSV  fa20a295-e75b-4ac2-b00e-65bb43626faa  dhs-data-for-rwanda   \n",
       "\n",
       "  dataset_org_title       dataset_last_modified  \\\n",
       "0   The DHS Program  2020-03-20T23:05:25.396054   \n",
       "1   The DHS Program  2020-03-20T23:05:25.396054   \n",
       "2   The DHS Program  2020-03-20T23:05:25.396054   \n",
       "3   The DHS Program  2020-03-20T23:05:25.396054   \n",
       "4   The DHS Program  2020-03-20T23:05:25.396054   \n",
       "\n",
       "                                        dataset_tags  \\\n",
       "0  [{'display_name': 'demographics', 'id': '7aa60...   \n",
       "1  [{'display_name': 'demographics', 'id': '7aa60...   \n",
       "2  [{'display_name': 'demographics', 'id': '7aa60...   \n",
       "3  [{'display_name': 'demographics', 'id': '7aa60...   \n",
       "4  [{'display_name': 'demographics', 'id': '7aa60...   \n",
       "\n",
       "                                      dataset_groups  \\\n",
       "0  [{'description': '', 'display_name': 'Rwanda',...   \n",
       "1  [{'description': '', 'display_name': 'Rwanda',...   \n",
       "2  [{'description': '', 'display_name': 'Rwanda',...   \n",
       "3  [{'description': '', 'display_name': 'Rwanda',...   \n",
       "4  [{'description': '', 'display_name': 'Rwanda',...   \n",
       "\n",
       "   dataset_total_res_downloads  ...  type  \\\n",
       "0                         5480  ...   csv   \n",
       "1                         5480  ...   csv   \n",
       "2                         5480  ...   csv   \n",
       "3                         5480  ...   csv   \n",
       "4                         5480  ...   csv   \n",
       "\n",
       "                                             dataset sheet  \\\n",
       "0  {'archived': False, 'batch': '6e71bebe-0a04-4c...         \n",
       "1  {'archived': False, 'batch': '6e71bebe-0a04-4c...         \n",
       "2  {'archived': False, 'batch': '6e71bebe-0a04-4c...         \n",
       "3  {'archived': False, 'batch': '6e71bebe-0a04-4c...         \n",
       "4  {'archived': False, 'batch': '6e71bebe-0a04-4c...         \n",
       "\n",
       "                                            resource num_rows num_cols  \\\n",
       "0  {'alt_url': 'https://data.humdata.org/dataset/...      108       25   \n",
       "1  {'alt_url': 'https://data.humdata.org/dataset/...       61       25   \n",
       "2  {'alt_url': 'https://data.humdata.org/dataset/...       94       27   \n",
       "3  {'alt_url': 'https://data.humdata.org/dataset/...      179       27   \n",
       "4  {'alt_url': 'https://data.humdata.org/dataset/...      118       27   \n",
       "\n",
       "                                             columns  \\\n",
       "0  ['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...   \n",
       "1  ['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...   \n",
       "2  ['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...   \n",
       "3  ['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...   \n",
       "4  ['ISO3', 'DataId', 'Indicator', 'Value', 'Prec...   \n",
       "\n",
       "                                           first_row has_hxl_header  \\\n",
       "0  ['#country+code', '#meta+id', '#indicator+name...           True   \n",
       "1  ['#country+code', '#meta+id', '#indicator+name...           True   \n",
       "2  ['#country+code', '#meta+id', '#indicator+name...           True   \n",
       "3  ['#country+code', '#meta+id', '#indicator+name...           True   \n",
       "4  ['#country+code', '#meta+id', '#indicator+name...           True   \n",
       "\n",
       "                                         first_nrows  \n",
       "0              ISO3    DataId                    ...  \n",
       "1              ISO3    DataId                    ...  \n",
       "2              ISO3    DataId                    ...  \n",
       "3              ISO3    DataId                    ...  \n",
       "4              ISO3    DataId                    ...  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "hxl_resources_data_details = pd.read_pickle(f\"./data/hxl_data_details.pkl\")\n",
    "display(hxl_resources_data_details.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(25695, 22)\n"
     ]
    }
   ],
   "source": [
    "print(hxl_resources_data_details.shape)"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Create some training data\n",
    "\n",
    "We need to create training data where every row relates to a column in a datafile, for which we need HXL tags. \n",
    "\n",
    "When I first tried this, I downloaded data, created by the data then did a typical test/train split. However, I noticed that columns from resources (files) in the same dataset could occur in both training and test. I felt this an unrealistic test, because when we predict in the wild it will be for a dataset that likely wasn't in the traning data. So being cautious I made the test/train split by ...\n",
    "\n",
    "1. Create a unique list of datasets (Note: In HDX a dataset can hold multiple 'resources', ie data files)\n",
    "2. Used the classic scikit learn train_test_split to split dataset dataframe\n",
    "3. Then using the train and test datasets, I built corresponding train/test resources (files) datafraes\n",
    "4. Using thes resources train/test dataframes, I finally built the model train/test dataframes wiith a row per column\n",
    "\n",
    "Convoluted, I know, but I felt it was more realistic.\n",
    "\n",
    "#### Split HXL datasets into train/test, then use to to create train/test dataframes of resources (files) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 332,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train datasets: 2530\n",
      "Test datasets: 1085\n",
      "Train resources: 12970\n",
      "Test resources: 5713\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>resource_id</th>\n",
       "      <th>resource_name</th>\n",
       "      <th>resource_format</th>\n",
       "      <th>dataset_id</th>\n",
       "      <th>dataset_name</th>\n",
       "      <th>dataset_org_title</th>\n",
       "      <th>dataset_last_modified</th>\n",
       "      <th>dataset_tags</th>\n",
       "      <th>dataset_groups</th>\n",
       "      <th>dataset_total_res_downloads</th>\n",
       "      <th>...</th>\n",
       "      <th>type</th>\n",
       "      <th>dataset</th>\n",
       "      <th>sheet</th>\n",
       "      <th>resource</th>\n",
       "      <th>num_rows</th>\n",
       "      <th>num_cols</th>\n",
       "      <th>columns</th>\n",
       "      <th>first_row</th>\n",
       "      <th>has_hxl_header</th>\n",
       "      <th>first_nrows</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>c144ee8a-cbe6-4159-836a-79a87c4bcf63</td>\n",
       "      <td>Social Development Indicators for Bhutan</td>\n",
       "      <td>CSV</td>\n",
       "      <td>d0d61099-c466-499a-8989-1f898d095b03</td>\n",
       "      <td>world-bank-social-development-indicators-for-b...</td>\n",
       "      <td>World Bank Group</td>\n",
       "      <td>2022-12-27T13:50:41.047263</td>\n",
       "      <td>[{'display_name': 'development', 'id': '3be50d...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Bhutan',...</td>\n",
       "      <td>401</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '0cb56400-6f98-4b...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>793</td>\n",
       "      <td>6</td>\n",
       "      <td>['Country Name', 'Country ISO3', 'Year', 'Indi...</td>\n",
       "      <td>['#country+name', '#country+code', '#date+year...</td>\n",
       "      <td>True</td>\n",
       "      <td>Country Name   Country ISO3        Year  \\...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>68</th>\n",
       "      <td>d1e64e90-b78d-4d6c-9d7f-d9ca9351b280</td>\n",
       "      <td>QuickCharts-Social Development Indicators for ...</td>\n",
       "      <td>CSV</td>\n",
       "      <td>d0d61099-c466-499a-8989-1f898d095b03</td>\n",
       "      <td>world-bank-social-development-indicators-for-b...</td>\n",
       "      <td>World Bank Group</td>\n",
       "      <td>2022-12-27T13:50:41.047263</td>\n",
       "      <td>[{'display_name': 'development', 'id': '3be50d...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Bhutan',...</td>\n",
       "      <td>401</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '0cb56400-6f98-4b...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>117</td>\n",
       "      <td>4</td>\n",
       "      <td>['Country ISO3', 'Year', 'Indicator Code', 'Va...</td>\n",
       "      <td>['#country+code', '#date+year', '#indicator+co...</td>\n",
       "      <td>True</td>\n",
       "      <td>Country ISO3        Year   Indicator Code ...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>92721633-d31d-4d71-97fe-526eb28ade08</td>\n",
       "      <td>Conflict Data for Angola</td>\n",
       "      <td>CSV</td>\n",
       "      <td>149215f8-e978-48fa-a843-d0e186d22821</td>\n",
       "      <td>ucdp-data-for-angola</td>\n",
       "      <td>HDX</td>\n",
       "      <td>2023-01-03T23:31:55.486746</td>\n",
       "      <td>[{'display_name': 'hxl', 'id': 'a0fbb23a-6aad-...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Angola',...</td>\n",
       "      <td>518</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '6e2477a8-3d2f-41...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>2079</td>\n",
       "      <td>49</td>\n",
       "      <td>['id', 'relid', 'year', 'active_year', 'code_s...</td>\n",
       "      <td>[nan, nan, '#date+year', nan, nan, nan, nan, n...</td>\n",
       "      <td>True</td>\n",
       "      <td>id             relid        year  acti...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>71</th>\n",
       "      <td>fe8a84a7-fea1-4935-8307-e7784cba2c23</td>\n",
       "      <td>List of airports in Niue (HXL tags)</td>\n",
       "      <td>CSV</td>\n",
       "      <td>9805334d-9293-40a5-a7c4-e5cd04acd8f0</td>\n",
       "      <td>ourairports-niu</td>\n",
       "      <td>OurAirports</td>\n",
       "      <td>2019-11-07T23:01:27.943973</td>\n",
       "      <td>[{'display_name': 'airports', 'id': '9c637e47-...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Niue', '...</td>\n",
       "      <td>0</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': 'a99c38b8-af33-41...</td>\n",
       "      <td></td>\n",
       "      <td>{'cache_last_updated': None, 'cache_url': None...</td>\n",
       "      <td>2</td>\n",
       "      <td>23</td>\n",
       "      <td>['id', 'ident', 'type', 'name', 'latitude_deg'...</td>\n",
       "      <td>['#meta +id', '#meta +code', '#loc +airport +t...</td>\n",
       "      <td>True</td>\n",
       "      <td>id        ident                 type...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>73</th>\n",
       "      <td>68951a4d-883f-4e0b-b976-41ea75d14e3a</td>\n",
       "      <td>Climate Change Indicators for Brazil</td>\n",
       "      <td>CSV</td>\n",
       "      <td>dabf9f3c-bd68-4515-8c1d-2b0c6af171e5</td>\n",
       "      <td>world-bank-climate-change-indicators-for-brazil</td>\n",
       "      <td>World Bank Group</td>\n",
       "      <td>2022-12-27T13:43:51.576025</td>\n",
       "      <td>[{'display_name': 'hxl', 'id': 'a0fbb23a-6aad-...</td>\n",
       "      <td>[{'description': '', 'display_name': 'Brazil',...</td>\n",
       "      <td>690</td>\n",
       "      <td>...</td>\n",
       "      <td>csv</td>\n",
       "      <td>{'archived': False, 'batch': '0cb56400-6f98-4b...</td>\n",
       "      <td></td>\n",
       "      <td>{'alt_url': 'https://data.humdata.org/dataset/...</td>\n",
       "      <td>1918</td>\n",
       "      <td>6</td>\n",
       "      <td>['Country Name', 'Country ISO3', 'Year', 'Indi...</td>\n",
       "      <td>['#country+name', '#country+code', '#date+year...</td>\n",
       "      <td>True</td>\n",
       "      <td>Country Name   Country ISO3        Year   ...</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 22 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                             resource_id  \\\n",
       "67  c144ee8a-cbe6-4159-836a-79a87c4bcf63   \n",
       "68  d1e64e90-b78d-4d6c-9d7f-d9ca9351b280   \n",
       "70  92721633-d31d-4d71-97fe-526eb28ade08   \n",
       "71  fe8a84a7-fea1-4935-8307-e7784cba2c23   \n",
       "73  68951a4d-883f-4e0b-b976-41ea75d14e3a   \n",
       "\n",
       "                                        resource_name resource_format  \\\n",
       "67           Social Development Indicators for Bhutan             CSV   \n",
       "68  QuickCharts-Social Development Indicators for ...             CSV   \n",
       "70                           Conflict Data for Angola             CSV   \n",
       "71                List of airports in Niue (HXL tags)             CSV   \n",
       "73               Climate Change Indicators for Brazil             CSV   \n",
       "\n",
       "                              dataset_id  \\\n",
       "67  d0d61099-c466-499a-8989-1f898d095b03   \n",
       "68  d0d61099-c466-499a-8989-1f898d095b03   \n",
       "70  149215f8-e978-48fa-a843-d0e186d22821   \n",
       "71  9805334d-9293-40a5-a7c4-e5cd04acd8f0   \n",
       "73  dabf9f3c-bd68-4515-8c1d-2b0c6af171e5   \n",
       "\n",
       "                                         dataset_name dataset_org_title  \\\n",
       "67  world-bank-social-development-indicators-for-b...  World Bank Group   \n",
       "68  world-bank-social-development-indicators-for-b...  World Bank Group   \n",
       "70                               ucdp-data-for-angola               HDX   \n",
       "71                                    ourairports-niu       OurAirports   \n",
       "73    world-bank-climate-change-indicators-for-brazil  World Bank Group   \n",
       "\n",
       "         dataset_last_modified  \\\n",
       "67  2022-12-27T13:50:41.047263   \n",
       "68  2022-12-27T13:50:41.047263   \n",
       "70  2023-01-03T23:31:55.486746   \n",
       "71  2019-11-07T23:01:27.943973   \n",
       "73  2022-12-27T13:43:51.576025   \n",
       "\n",
       "                                         dataset_tags  \\\n",
       "67  [{'display_name': 'development', 'id': '3be50d...   \n",
       "68  [{'display_name': 'development', 'id': '3be50d...   \n",
       "70  [{'display_name': 'hxl', 'id': 'a0fbb23a-6aad-...   \n",
       "71  [{'display_name': 'airports', 'id': '9c637e47-...   \n",
       "73  [{'display_name': 'hxl', 'id': 'a0fbb23a-6aad-...   \n",
       "\n",
       "                                       dataset_groups  \\\n",
       "67  [{'description': '', 'display_name': 'Bhutan',...   \n",
       "68  [{'description': '', 'display_name': 'Bhutan',...   \n",
       "70  [{'description': '', 'display_name': 'Angola',...   \n",
       "71  [{'description': '', 'display_name': 'Niue', '...   \n",
       "73  [{'description': '', 'display_name': 'Brazil',...   \n",
       "\n",
       "    dataset_total_res_downloads  ...  type  \\\n",
       "67                          401  ...   csv   \n",
       "68                          401  ...   csv   \n",
       "70                          518  ...   csv   \n",
       "71                            0  ...   csv   \n",
       "73                          690  ...   csv   \n",
       "\n",
       "                                              dataset sheet  \\\n",
       "67  {'archived': False, 'batch': '0cb56400-6f98-4b...         \n",
       "68  {'archived': False, 'batch': '0cb56400-6f98-4b...         \n",
       "70  {'archived': False, 'batch': '6e2477a8-3d2f-41...         \n",
       "71  {'archived': False, 'batch': 'a99c38b8-af33-41...         \n",
       "73  {'archived': False, 'batch': '0cb56400-6f98-4b...         \n",
       "\n",
       "                                             resource num_rows num_cols  \\\n",
       "67  {'alt_url': 'https://data.humdata.org/dataset/...      793        6   \n",
       "68  {'alt_url': 'https://data.humdata.org/dataset/...      117        4   \n",
       "70  {'alt_url': 'https://data.humdata.org/dataset/...     2079       49   \n",
       "71  {'cache_last_updated': None, 'cache_url': None...        2       23   \n",
       "73  {'alt_url': 'https://data.humdata.org/dataset/...     1918        6   \n",
       "\n",
       "                                              columns  \\\n",
       "67  ['Country Name', 'Country ISO3', 'Year', 'Indi...   \n",
       "68  ['Country ISO3', 'Year', 'Indicator Code', 'Va...   \n",
       "70  ['id', 'relid', 'year', 'active_year', 'code_s...   \n",
       "71  ['id', 'ident', 'type', 'name', 'latitude_deg'...   \n",
       "73  ['Country Name', 'Country ISO3', 'Year', 'Indi...   \n",
       "\n",
       "                                            first_row has_hxl_header  \\\n",
       "67  ['#country+name', '#country+code', '#date+year...           True   \n",
       "68  ['#country+code', '#date+year', '#indicator+co...           True   \n",
       "70  [nan, nan, '#date+year', nan, nan, nan, nan, n...           True   \n",
       "71  ['#meta +id', '#meta +code', '#loc +airport +t...           True   \n",
       "73  ['#country+name', '#country+code', '#date+year...           True   \n",
       "\n",
       "                                          first_nrows  \n",
       "67      Country Name   Country ISO3        Year  \\...  \n",
       "68      Country ISO3        Year   Indicator Code ...  \n",
       "70          id             relid        year  acti...  \n",
       "71            id        ident                 type...  \n",
       "73      Country Name   Country ISO3        Year   ...  \n",
       "\n",
       "[5 rows x 22 columns]"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "hdx_datasets = []\n",
    "for row, data in hxl_resources_data_details.iterrows():\n",
    "    d = {\"name\": data[\"dataset\"][\"title\"], \"id\": data[\"dataset\"][\"id\"]}\n",
    "    hdx_datasets.append(d)\n",
    "\n",
    "hdx_datasets = pd.DataFrame(hdx_datasets)\n",
    "hdx_datasets = hdx_datasets.drop_duplicates(subset=[\"id\"])\n",
    "\n",
    "# Train test split on datasets to avoid having very similar columns in train and test\n",
    "X_train_datasets, X_test_datasets = train_test_split(\n",
    "    hdx_datasets, test_size=0.3, random_state=42\n",
    ")\n",
    "\n",
    "print(f\"Train datasets: {len(X_train_datasets)}\")\n",
    "print(f\"Test datasets: {len(X_test_datasets)}\")\n",
    "\n",
    "X_train_resources = []\n",
    "X_test_resources = []\n",
    "for row, data in hxl_resources_data_details.iterrows():\n",
    "    if data[\"dataset\"][\"id\"] in X_train_datasets[\"id\"].values:\n",
    "        X_train_resources.append(data)\n",
    "    else:\n",
    "        X_test_resources.append(data)\n",
    "\n",
    "X_train_resources = pd.DataFrame(X_train_resources)\n",
    "X_test_resources = pd.DataFrame(X_test_resources)\n",
    "\n",
    "# Now lets prune resources which don't have HDX headers. Even though our search was for 'HXL', not all resources have HXL headers\n",
    "X_train_resources = X_train_resources.loc[X_train_resources[\"has_hxl_header\"] == True]\n",
    "X_test_resources = X_test_resources.loc[X_test_resources[\"has_hxl_header\"] == True]\n",
    "\n",
    "# To be similiar with previous prediction studies, we'll only use CSV which are generally cleaner on HDX because they\n",
    "# don't have a lot of random tabs  \n",
    "X_train_resources = X_train_resources.loc[X_train_resources[\"resource_format\"] == \"CSV\"]\n",
    "X_test_resources = X_test_resources.loc[X_test_resources[\"resource_format\"] == \"CSV\"]\n",
    "\n",
    "print(f\"Train resources: {len(X_train_resources)}\")\n",
    "print(f\"Test resources: {len(X_test_resources)}\")\n",
    "\n",
    "display(X_train_resources.head())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 334,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Check that there are no dataset id's in the training set which are also in the test set\n",
    "test_dataset_ids = X_test_resources[\"dataset_id\"].values\n",
    "train_dataset_ids = X_train_resources[\"dataset_id\"].values\n",
    "\n",
    "for id in test_dataset_ids:\n",
    "    if id in train_dataset_ids:\n",
    "        print(f\"Dataset {id} is in both train and test\")\n",
    "        sys.exit()\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Build model train/test dataframes (of columns) from train/test resources (files) list\n",
    "\n",
    "In order to prompt GPT-3 I combine the column name and 1st 10 rows of data in that column. The response, or 'completion', is set to the HXL tag and attributes. Here is an example ...\n",
    "\n",
    "`{\"prompt\": \"  'scheduled_service' | \\\"['1', '1', '0', '0', '0', '0', '0', '0']\\\"\", \"completion\": \" #status+scheduled\"}`\n",
    "\n",
    "The format for GPT-3 is very particular and took a while to get right! Things like having a space at the start of completion are recommended by OpenAI."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_prompt(col_name, data):\n",
    "    \"\"\"\n",
    "    This function takes column name and sample data and create the prompt for GPT-3.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    col_name : str\n",
    "        Column name\n",
    "    data : list\n",
    "        A list of sample data\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    prompt : string\n",
    "        A prompt for GPT-3.\n",
    "    \"\"\"\n",
    "    ld = len(data) - 1\n",
    "    col_data = json.dumps(str(list(data.iloc[1:ld])))\n",
    "    prompt = f\" {col_name} | {col_data}\".lower()\n",
    "    return prompt\n",
    "   \n",
    "\n",
    "def create_training_set(resources):\n",
    "    \"\"\"\n",
    "    Builds a jsonl training data file for GPT-3 where each row is a prompt for a column HXL tag.\n",
    "\n",
    "    It will only output prompts where the sample data for the column didn't contain nans.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    resources : pandas dataframe\n",
    "        A dataframe of HDX resources\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    train_data : list\n",
    "        A list of prompts and completions for the HXL tag autocomplete feature.\n",
    "    \"\"\"\n",
    "    train_data = []\n",
    "    for row, d in resources.iterrows():\n",
    "        if d[\"has_hxl_header\"] == True:\n",
    "            cols = d[\"columns\"][1:-1].split(\",\")\n",
    "            hdxs = d[\"first_row\"][1:-1].split(\",\")\n",
    "            data = d[\"first_nrows\"]\n",
    "            has_hxl_header = d[\"has_hxl_header\"]\n",
    "            if len(cols) == len(hdxs) and len(cols) > 1:\n",
    "                ld = len(data) - 1\n",
    "                for i in range(0, len(cols)):\n",
    "                    if i < len(hdxs):\n",
    "                        hdx = re.sub(\"'|\\\"\", \"\", hdxs[i])\n",
    "                    # Only include is has HXL tags and good sample data in column\n",
    "                    if has_hxl_header == True and hdx != np.nan:\n",
    "                        prompt = get_prompt(cols[i], data.iloc[:,i])\n",
    "                        if 'nan' not in hdx and 'nan, nan' not in prompt:\n",
    "                            p = {\n",
    "                                \"prompt\": prompt,\n",
    "                                \"completion\": f\" {hdx}\",\n",
    "                            }\n",
    "                            train_data.append(p)\n",
    "    return train_data\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['event', 'displaced', 'i_ky', 'funder', '#name', 'value', 'i_ru', 'age_95_99', '#indicator', 'keywords', 'stock', '#value', 'market', 'v_pcode', '#valid_datatype', 'start', 'text', 'external', 'infected', 'first']\n"
     ]
    }
   ],
   "source": [
    "def get_hxl_tags_list(resources):\n",
    "    \"\"\"\n",
    "    This function takes a dataframe of resources and returns a master list of HXL tags.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    resources : pandas dataframe\n",
    "        A dataframe of HDX resources\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    hxl_tags : list\n",
    "        A list of HXL tags.\n",
    "    \"\"\"\n",
    "    hxl_tags = []\n",
    "    for row, d in resources.iterrows():\n",
    "        if d[\"has_hxl_header\"] == True:\n",
    "            fr = d[\"first_row\"].replace(\" \", \"\")\n",
    "            for c in fr.split(\",\"):\n",
    "                fr = re.sub(\"\\[|\\]|\\\"|\\'\",\"\", c)\n",
    "                hdxs = fr.split(\"+\")\n",
    "                for h in hdxs:\n",
    "                    if h not in hxl_tags and len(h) > 0:\n",
    "                        hxl_tags.append(h.lower())\n",
    "    hxl_tags = list(set(hxl_tags))\n",
    "    hxl_tags.remove('nan')\n",
    "    return hxl_tags\n",
    "\n",
    "# Extract unique list of tags and attributes as seen in training data\n",
    "hxl_supported_tags = get_hxl_tags_list(X_train_resources)\n",
    "print(hxl_supported_tags[0:20])\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training records: 136107\n",
      "Done\n"
     ]
    }
   ],
   "source": [
    "# Create training data\n",
    "X_train = create_training_set(X_train_resources)\n",
    "print(f\"Training records: {len(X_train)}\")\n",
    "\n",
    "train_file = \"fine_tune_openai_train_just_csv.jsonl\"\n",
    "\n",
    "with open(train_file, \"w\") as f:\n",
    "    for p in X_train:\n",
    "        json.dump(p, f)\n",
    "        f.write(\"\\n\")\n",
    "\n",
    "print(\"Done\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Fine tune GPT-3 using our HXL tag prompts\n",
    "\n",
    "OK, let's [fine tune GPT-3](https://beta.openai.com/docs/guides/fine-tuning) ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 179,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uploading training file ...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Upload progress: 100%|██████████| 28.5M/28.5M [00:00<00:00, 7.93Git/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Uploaded file from fine_tune_openai_train_just_csv.jsonl: file-WVlZgsTrzFut92lV5sBKZfjR\n",
      "Fine-tuning model ...\n",
      "Fine-tunning model with jobID: ft-i2QSzQNbbFOYcSYhntOI728H.\n"
     ]
    }
   ],
   "source": [
    "print(\"Uploading training file ...\")\n",
    "training_id = cli.FineTune._get_or_upload(train_file, True)\n",
    "# validation_id = cli.FineTune._get_or_upload(validation_file_name, True)\n",
    "\n",
    "print(\"Fine-tuning model ...\")\n",
    "create_args = {\n",
    "    \"training_file\": training_id,\n",
    "    # \"validation_file\": test_file,\n",
    "    \"model\": \"ada\",\n",
    "}\n",
    "# https://beta.openai.com/docs/api-reference/fine-tunes/create\n",
    "resp = openai.FineTune.create(**create_args)\n",
    "job_id = resp[\"id\"]\n",
    "status = resp[\"status\"]\n",
    "\n",
    "print(f\"Fine-tunning model with jobID: {job_id}.\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In the above I opted for a simple submit, got a cup of tea then came back and periodically ran the following pane (took about 2 hours for 156k records, each about 20-30 tokens in prompt and completions). You can also set up a stream to monitor automatically, some examples [here](https://github.com/openai/openai-cookbook/blob/main/examples/azure/finetuning.ipynb) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#print(openai.FineTune.list())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "succeeded\n",
      "{\n",
      "  \"created_at\": 1674046907,\n",
      "  \"events\": [\n",
      "    {\n",
      "      \"created_at\": 1674046907,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Created fine-tune: ft-i2QSzQNbbFOYcSYhntOI728H\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674046950,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune costs $13.75\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674046951,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune enqueued. Queue number: 3\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674047270,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune is in the queue. Queue number: 2\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674047270,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune is in the queue. Queue number: 1\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674047334,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune is in the queue. Queue number: 0\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674047428,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune started\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674048171,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Completed epoch 1/4\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674048895,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Completed epoch 2/4\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674049614,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Completed epoch 3/4\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674050333,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Completed epoch 4/4\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674050352,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Uploaded model: ada:ft-personal-2023-01-18-13-59-12\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674050353,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Uploaded result file: file-c8cr0Qwlpxt5wYZM6ssuUk5W\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    },\n",
      "    {\n",
      "      \"created_at\": 1674050353,\n",
      "      \"level\": \"info\",\n",
      "      \"message\": \"Fine-tune succeeded\",\n",
      "      \"object\": \"fine-tune-event\"\n",
      "    }\n",
      "  ],\n",
      "  \"fine_tuned_model\": \"ada:ft-personal-2023-01-18-13-59-12\",\n",
      "  \"hyperparams\": {\n",
      "    \"batch_size\": 256,\n",
      "    \"learning_rate_multiplier\": 0.2,\n",
      "    \"n_epochs\": 4,\n",
      "    \"prompt_loss_weight\": 0.01\n",
      "  },\n",
      "  \"id\": \"ft-i2QSzQNbbFOYcSYhntOI728H\",\n",
      "  \"model\": \"ada\",\n",
      "  \"object\": \"fine-tune\",\n",
      "  \"organization_id\": \"org-yApYP2cR2HMTkNZxXMvP76FK\",\n",
      "  \"result_files\": [\n",
      "    {\n",
      "      \"bytes\": 128261,\n",
      "      \"created_at\": 1674050353,\n",
      "      \"filename\": \"compiled_results.csv\",\n",
      "      \"id\": \"file-c8cr0Qwlpxt5wYZM6ssuUk5W\",\n",
      "      \"object\": \"file\",\n",
      "      \"purpose\": \"fine-tune-results\",\n",
      "      \"status\": \"processed\",\n",
      "      \"status_details\": null\n",
      "    }\n",
      "  ],\n",
      "  \"status\": \"succeeded\",\n",
      "  \"training_files\": [\n",
      "    {\n",
      "      \"bytes\": 28493931,\n",
      "      \"created_at\": 1674046907,\n",
      "      \"filename\": \"fine_tune_openai_train_just_csv.jsonl\",\n",
      "      \"id\": \"file-WVlZgsTrzFut92lV5sBKZfjR\",\n",
      "      \"object\": \"file\",\n",
      "      \"purpose\": \"fine-tune\",\n",
      "      \"status\": \"processed\",\n",
      "      \"status_details\": null\n",
      "    }\n",
      "  ],\n",
      "  \"updated_at\": 1674050353,\n",
      "  \"validation_files\": []\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "result = openai.FineTune.retrieve(id=job_id)\n",
    "print(result['status'])\n",
    "print(result)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model run complete, here is your model: ada:ft-personal-2023-01-18-13-59-12\n"
     ]
    }
   ],
   "source": [
    "# Once model is done, this should return model for inference\n",
    "if result['status'] == 'succeeded':\n",
    "    model = result[\"fine_tuned_model\"]\n",
    "    print(f\"Model run complete, here is your model: {model}\")\n",
    "else:\n",
    "    print(result['status'])\n",
    "    print(\"Model not ready yet, run the above cell again in a bit\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Making predictions with GPT-3\n",
    "\n",
    "OK, we have a fine tuned model id, we can now use it to make predictions ...\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 325,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ada:ft-personal-2023-01-18-13-59-12\n"
     ]
    }
   ],
   "source": [
    "#model ='ada:ft-personal-2023-01-16-22-15-09  # All resources, no lower-case\n",
    "#model = 'ada:ft-personal-2021-01-16-22-15-09' # Just lower-case, CSV resources only\n",
    "print(model)\n",
    "\n",
    "\n",
    "def create_prediction_dataset_from_resources(resources):\n",
    "    \"\"\"\n",
    "    Generate a list of model column-level prompts from a list of resources (tables).\n",
    "\n",
    "    It will only output prompts where the sample data for the column didn't contain nans.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    resources : list\n",
    "        A list of dictionaries containing the resource name, columns, first_row, and first_nrows.\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    prediction_data : list\n",
    "        A list of dictionaries containing the prompts (one per column in resource table)\n",
    "    \"\"\"\n",
    "\n",
    "    prediction_data = []\n",
    "    for index, d in resources.iterrows():\n",
    "        cols = d[\"columns\"][1:-1].split(\",\")\n",
    "        hdxs = d[\"first_row\"][1:-1].split(\",\")\n",
    "        data = d[\"first_nrows\"]\n",
    "        has_hxl_header = d[\"has_hxl_header\"]\n",
    "        if len(cols) == len(hdxs) and len(cols) > 1:\n",
    "            ld = len(data) - 1\n",
    "            # Loop through columns \n",
    "            for i in range(0, len(cols)):\n",
    "                if i < len(hdxs) and i < data.shape[1]:\n",
    "                    prompt = get_prompt(cols[i], data.iloc[:,i])\n",
    "                    # Skip any prompts with at least two nan values in sample data\n",
    "                    if 'nan, nan' not in prompt:\n",
    "                        r = {\n",
    "                                \"prompt\": prompt\n",
    "                            }\n",
    "                        # If we were called with HXL tags (ie for test set), populate 'expected'\n",
    "                        if has_hxl_header == True:\n",
    "                            hdx = re.sub(\"'|\\\"| \", \"\", hdxs[i])\n",
    "                            # Row has HXL tags, but this particular column doesn't have tags\n",
    "                            if hdx == 'nan':\n",
    "                                continue\n",
    "                            else:\n",
    "                                r[\"expected\"]= hdx\n",
    "                        prediction_data.append(r)\n",
    "    return prediction_data\n",
    "\n",
    "def make_gpt3_prediction(prompt, model, temperature=0.99, max_tokens=13):\n",
    "    \"\"\"\n",
    "    Wrapper to call GPT-3 to make a prediction (completion) on a single prompt.\n",
    "    Also calls post_process() to clean up the prediction.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    prompt : str\n",
    "        Prompt to use for prediction\n",
    "    model : str\n",
    "        GPT-3 model to use\n",
    "    temperature : float\n",
    "        Temperature to use for sampling\n",
    "    max_tokens : int\n",
    "        Maximum number of tokens to use for sampling\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    result : dict\n",
    "        Dictionary with prompt, predicted, predicted_post_processed\n",
    "    \"\"\"\n",
    "    result = {}\n",
    "    result[\"prompt\"] = prompt\n",
    "    model_result = openai.Completion.create(\n",
    "        engine=model,\n",
    "        prompt=prompt,\n",
    "        temperature=temperature,\n",
    "        max_tokens=max_tokens,\n",
    "        top_p=1,\n",
    "        frequency_penalty=0,\n",
    "        presence_penalty=0,\n",
    "        stop=[\"\\n\"],\n",
    "        logprobs=1\n",
    "    )\n",
    "    result[\"predicted\"] = model_result[\"choices\"][0][\"text\"].replace(\" \",\"\")\n",
    "    result[\"logprobs\"]  = model_result['choices'][0]['logprobs']['top_logprobs']\n",
    "    return result\n",
    "\n",
    "def make_gpt3_predictions(\n",
    "    sample_size, prediction_data, model, temperature=0.99, max_tokens=13, logprob_cutoff=-0.01\n",
    "):\n",
    "\n",
    "    \"\"\"\n",
    "    Wrapper to call GPT-3 to make predictions on test file for sample_size samples.\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    sample_size : int\n",
    "        Number of predictions to make from test file\n",
    "    prediction_data : list\n",
    "        List of dictionaries with prompts\n",
    "    model : str\n",
    "        GPT-3 model to use\n",
    "    postprocess : bool\n",
    "        Whether to postprocess the predictions\n",
    "    temperature : float\n",
    "        Temperature to use for sampling\n",
    "    max_tokens : int\n",
    "        Maximum number of tokens to use for sampling\n",
    "    prob_cutoff : float\n",
    "        Logprob cutoff for filtering out low probability tokens\n",
    "\n",
    "    Returns\n",
    "    -------\n",
    "    results : list\n",
    "        List of dictionaries with prompt, predicted, predicted_post_processed\n",
    "    \"\"\"\n",
    "    results = []\n",
    "    prediction_data = random.sample(prediction_data, sample_size)\n",
    "    for i in range(0, sample_size):\n",
    "        prompt = prediction_data[i][\"prompt\"]\n",
    "        res = make_gpt3_prediction(\n",
    "            prompt, model, temperature, max_tokens\n",
    "        )\n",
    "\n",
    "        # Filter out low logprob predictions\n",
    "        pred = \"\"\n",
    "        seen_tokens = []\n",
    "        for w in res[\"logprobs\"]:\n",
    "            token = list(w.keys())[0]\n",
    "            prob = w[token]\n",
    "            if prob > logprob_cutoff and token not in seen_tokens:\n",
    "                pred += token\n",
    "                if '+' not in token:\n",
    "                    seen_tokens.append(token)\n",
    "            else:\n",
    "                break\n",
    "        pred = re.sub(r\" |\\+$|\\+v_$\", \"\", pred)\n",
    "\n",
    "\n",
    "        r = {\n",
    "                \"prompt\": prompt,\n",
    "                \"predicted\": res[\"predicted\"],\n",
    "                \"predicted_log_prob_cutoff\": pred,\n",
    "                #\"logprobs\": res[\"logprobs\"]\n",
    "            }\n",
    "        # For test sets we have expected values, add back for performance reporting\n",
    "        if \"expected\" in prediction_data[i]:\n",
    "            r['expected'] = prediction_data[i]['expected'].replace(' ', '')\n",
    "        results.append(r)\n",
    "    return results\n",
    "\n",
    "\n",
    "def output_prediction_metrics(results, prediction_field=\"predicted_post_processed\"):\n",
    "    \"\"\"\n",
    "    Prints out model performance report if provided results in the format:\n",
    "\n",
    "    [\n",
    "        {\n",
    "            'prompt': ' \\'ISO3\\' | \"[\\'RWA\\', \\'RWA\\', \\'RWA\\', \\'RWA\\', \\'RWA\\', \\'RWA\\', \\'RWA\\', \\'RWA\\']\"', \n",
    "            'predicted': ' #country+code+iso3+v_iso3+', \n",
    "            'predicted_post_processed': '#country+code', \n",
    "            'expected': '#country+code'\n",
    "        }, \n",
    "        ... etc ...\n",
    "    ]\n",
    "\n",
    "    Parameters\n",
    "    ----------\n",
    "    results : list\n",
    "        See above for format\n",
    "    prediction_field : str\n",
    "        Field name of element with prediction. Handy for comparing raw and post-processed predictions.\n",
    "    \"\"\"\n",
    "    y_test = []\n",
    "    y_pred = []\n",
    "    y_justtag_test = []\n",
    "    y_justtag_pred = []\n",
    "    for r in results:\n",
    "        if \"expected\" not in r:\n",
    "            print(\"Provided results do not contain expected values.\")\n",
    "            sys.exit()\n",
    "        y_pred.append(r[prediction_field])\n",
    "        y_test.append(r[\"expected\"])\n",
    "        expected_tag = r[\"expected\"].split(\"+\")[0]\n",
    "        predicted_tag = r[prediction_field].split(\"+\")[0]\n",
    "        y_justtag_test.append(expected_tag)\n",
    "        y_justtag_pred.append(predicted_tag)\n",
    "\n",
    "    print(f\"GPT-3 results for {prediction_field}, {len(results)} predictions ...\")\n",
    "    print(\"\\nJust HXL tags ...\\n\")\n",
    "    print(f\"Accuracy: {round(accuracy_score(y_justtag_test, y_justtag_pred),2)}\")\n",
    "    print(\n",
    "        f\"Precision: {round(precision_score(y_justtag_test, y_justtag_pred, average='weighted', zero_division=0),2)}\"\n",
    "    )\n",
    "    print(\n",
    "        f\"Recall: {round(recall_score(y_justtag_test, y_justtag_pred, average='weighted', zero_division=0),2)}\"\n",
    "    )\n",
    "    print(\n",
    "        f\"F1: {round(f1_score(y_justtag_test, y_justtag_pred, average='weighted', zero_division=0),2)}\"\n",
    "    )\n",
    "\n",
    "    print(f\"\\nTags and attributes with {prediction_field} ...\\n\")\n",
    "    print(f\"Accuracy: {round(accuracy_score(y_test, y_pred),2)}\")\n",
    "    print(\n",
    "        f\"Precision: {round(precision_score(y_test, y_pred, average='weighted', zero_division=0),2)}\"\n",
    "    )\n",
    "    print(\n",
    "        f\"Recall: {round(recall_score(y_test, y_pred, average='weighted', zero_division=0),2)}\"\n",
    "    )\n",
    "    print(\n",
    "        f\"F1: {round(f1_score(y_test, y_pred, average='weighted', zero_division=0),2)}\"\n",
    "    )\n",
    "\n",
    "    return \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 348,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Building model input ...\n",
      "59657\n",
      "Making GPT-3 predictions (completions) ...\n",
      "500\n",
      "\n",
      "Model output example ...\n",
      "[\n",
      "  {\n",
      "    \"prompt\": \"  'funder_ref' | \\\"['xm-dac-47015', 'xi-iati-ebrd', 'ca_on-ont-1597880', 'xi-iati-ebrd', 'xi-iati-ebrd', 'gb-chc-209131', 'xi-iati-ebrd', 'xi-iati-ebrd']\\\"\",\n",
      "    \"predicted\": \"#org+funder+id+external+usd+Dough_IO+aids\",\n",
      "    \"predicted_log_prob_cutoff\": \"#org+funder+id\",\n",
      "    \"expected\": \"#org+funder+id\"\n",
      "  },\n",
      "  {\n",
      "    \"prompt\": \" 'gho (code)' | \\\"['cco_1', 'cco_2', 'cco_3', 'itu_ict_1', 'itu_ict_2', 'whs10_8', 'cco_2', 'cco_3']\\\"\",\n",
      "    \"predicted\": \"#indicator+code+v_indicator+bl_3_indicator+obstetric\",\n",
      "    \"predicted_log_prob_cutoff\": \"#indicator+code\",\n",
      "    \"expected\": \"#indicator+code\"\n",
      "  },\n",
      "  {\n",
      "    \"prompt\": \"  'destusageyearstart' | \\\"['2022', '2023', '2022', '2023', '2023', '2023', '2022', '2022']\\\"\",\n",
      "    \"predicted\": \"#date+year+start+impl+list+list+impl+list_xml+\",\n",
      "    \"predicted_log_prob_cutoff\": \"#date+year+start+impl\",\n",
      "    \"expected\": \"#date+year+start+impl\"\n",
      "  },\n",
      "  {\n",
      "    \"prompt\": \"  'publishstate (code)' | \\\"['published', 'published', 'published', 'published', 'published', 'published', 'published', 'published']\\\"\",\n",
      "    \"predicted\": \"#status+code+published+usd+v_imp2_usd+v\",\n",
      "    \"predicted_log_prob_cutoff\": \"#status+code\",\n",
      "    \"expected\": \"#status+code\"\n",
      "  },\n",
      "  {\n",
      "    \"prompt\": \"  'byvariableid' | \\\"['0', '0', '0', '0', '0', '0', '0', '0']\\\"\",\n",
      "    \"predicted\": \"#indicator+label+code+v_aterappeal+val_0_9\",\n",
      "    \"predicted_log_prob_cutoff\": \"#indicator+label+code\",\n",
      "    \"expected\": \"#indicator+label+code\"\n",
      "  }\n",
      "]\n",
      "Done\n"
     ]
    }
   ],
   "source": [
    "# Generate the prompts we want GPT-3 to complete\n",
    "print(\"Building model input ...\")\n",
    "\n",
    "prediction_data = create_prediction_dataset_from_resources(X_test_resources)\n",
    "\n",
    "print(len(prediction_data))\n",
    "with open(\"prediction_input_just_csv.json\", \"w\") as f:\n",
    "    json.dump(prediction_data, f, indent=4, sort_keys=False)\n",
    "\n",
    "# How many predictions to try from the test set\n",
    "sample_size = 500\n",
    "\n",
    "# Make the predictions\n",
    "print(\"Making GPT-3 predictions (completions) ...\")\n",
    "results = make_gpt3_predictions(\n",
    "   sample_size, prediction_data, model, temperature=0.99, max_tokens=20, logprob_cutoff=-0.001\n",
    ")\n",
    "\n",
    "print(len(results))\n",
    "\n",
    "with open(\"results_just_csv.json\", \"w\") as f:\n",
    "    json.dump(results, f, indent=4, sort_keys=False)\n",
    "\n",
    "print(\"\\nModel output example ...\")\n",
    "print(json.dumps(results[0:5], indent=2))\n",
    "\n",
    "print(\"Done\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### GPT-3 performance with no post-processing\n",
    "\n",
    "If we just take the model prediction as-is for 500 prompts, performance looks like this ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 349,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GPT-3 results for predicted, 500 predictions ...\n",
      "\n",
      "Just HXL tags ...\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 0.99\n",
      "Recall: 1.0\n",
      "F1: 0.99\n",
      "\n",
      "Tags and attributes with predicted ...\n",
      "\n",
      "Accuracy: 0.0\n",
      "Precision: 0.0\n",
      "Recall: 0.0\n",
      "F1: 0.0\n"
     ]
    }
   ],
   "source": [
    "output_prediction_metrics(results, prediction_field=\"predicted\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Wait what? That's weird, Very good performance on just tags, but 0 on tags and attributes. Let's look at the results ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 350,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "    \"prompt\": \"  'funder_ref' | \\\"['xm-dac-47015', 'xi-iati-ebrd', 'ca_on-ont-1597880', 'xi-iati-ebrd', 'xi-iati-ebrd', 'gb-chc-209131', 'xi-iati-ebrd', 'xi-iati-ebrd']\\\"\",\n",
      "    \"predicted\": \"#org+funder+id+external+usd+Dough_IO+aids\",\n",
      "    \"expected\": \"#org+funder+id\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \" 'gho (code)' | \\\"['cco_1', 'cco_2', 'cco_3', 'itu_ict_1', 'itu_ict_2', 'whs10_8', 'cco_2', 'cco_3']\\\"\",\n",
      "    \"predicted\": \"#indicator+code+v_indicator+bl_3_indicator+obstetric\",\n",
      "    \"expected\": \"#indicator+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'destusageyearstart' | \\\"['2022', '2023', '2022', '2023', '2023', '2023', '2022', '2022']\\\"\",\n",
      "    \"predicted\": \"#date+year+start+impl+list+list+impl+list_xml+\",\n",
      "    \"expected\": \"#date+year+start+impl\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'publishstate (code)' | \\\"['published', 'published', 'published', 'published', 'published', 'published', 'published', 'published']\\\"\",\n",
      "    \"predicted\": \"#status+code+published+usd+v_imp2_usd+v\",\n",
      "    \"expected\": \"#status+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'byvariableid' | \\\"['0', '0', '0', '0', '0', '0', '0', '0']\\\"\",\n",
      "    \"predicted\": \"#indicator+label+code+v_aterappeal+val_0_9\",\n",
      "    \"expected\": \"#indicator+label+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'precision' | \\\"['1', '1', '1', '1', '1', '1', '0', '0']\\\"\",\n",
      "    \"predicted\": \"#indicator+precision+code+v_precision+code_precision_\",\n",
      "    \"expected\": \"#indicator+precision\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'indicator code' | \\\"['ic.frm.femm.zs', 'ic.frm.femm.zs', 'ic.frm.femo.zs', 'ic.frm.femo.zs', 'se.adt.1524.lt.fe.zs', 'se.adt.1524.lt.fe.zs', 'se.adt.1524.lt.fe.zs', 'se.adt.1524.lt.fe.zs']\\\"\",\n",
      "    \"predicted\": \"#indicator+code+ind+list+en_enrich_femo_se\",\n",
      "    \"expected\": \"#indicator+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \" 'iso3' | \\\"['npl', 'npl', 'npl', 'npl', 'npl', 'npl', 'npl', 'npl']\\\"\",\n",
      "    \"predicted\": \"#country+code+ISO3+v_iso3_nn_iso4_v_\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'year' | \\\"['2018', '2017', '2016', '2015', '2014', '2013', '2012', '2011']\\\"\",\n",
      "    \"predicted\": \"#date+year+playground+flag+year+playground+ip_a_\",\n",
      "    \"expected\": \"#date+year\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \" 'country name' | \\\"['bahrain', 'bahrain', 'bahrain', 'bahrain', 'bahrain', 'bahrain', 'bahrain', 'bahrain']\\\"\",\n",
      "    \"predicted\": \"#country+name+v_iso3+country_fficage+pct_iso3\",\n",
      "    \"expected\": \"#country+name\"\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "for r in results[0:10]:\n",
    "    if r[\"predicted\"] != r[\"expected\"]:\n",
    "        r = {\n",
    "            \"prompt\": r[\"prompt\"],\n",
    "            \"predicted\": r[\"predicted\"],\n",
    "            \"expected\": r[\"expected\"]\n",
    "        }\n",
    "        print(json.dumps(r, indent=4, sort_keys=False))\n",
    "\n"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's see how often the expected tags and attributes occurred in the first half of the predictions ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 351,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      " Out of 500 predictions, the expected tags and attributes where in the predicted tags and attributes 99.4% of the time.\n"
     ]
    }
   ],
   "source": [
    "\n",
    "passes = 0\n",
    "fails = 0\n",
    "for r in results:\n",
    "    if r[\"predicted\"].startswith(r[\"expected\"]):\n",
    "        passes += 1\n",
    "    else:\n",
    "        fails += 1\n",
    "        #print(json.dumps(r, indent=4, sort_keys=False))\n",
    "\n",
    "print(f\" Out of {passes + fails} predictions, the expected tags and attributes where in the predicted tags and attributes {round(100*passes/(passes+fails),1)}% of the time.\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "So GPT-3 has great accuracy for predicting tags and attributes (basically the expected values are almost always the first part of the prediction), but adds extra attributes at the end.\n",
    "\n",
    "So, how to exclude those extra tokens?\n",
    "\n",
    "Well, it runs out that the GPT-3 API returns log probabilities for each token. As you will notice above, we also calculated a prediction assuming a cutoff for this. Let's see how that performed .."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 352,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GPT-3 results for predicted_log_prob_cutoff, 500 predictions ...\n",
      "\n",
      "Just HXL tags ...\n",
      "\n",
      "Accuracy: 0.99\n",
      "Precision: 1.0\n",
      "Recall: 0.99\n",
      "F1: 0.99\n",
      "\n",
      "Tags and attributes with predicted_log_prob_cutoff ...\n",
      "\n",
      "Accuracy: 0.94\n",
      "Precision: 0.99\n",
      "Recall: 0.94\n",
      "F1: 0.95\n"
     ]
    }
   ],
   "source": [
    "output_prediction_metrics(results, prediction_field=\"predicted_log_prob_cutoff\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Much better! Let's see where it's having problems ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 345,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{\n",
      "    \"prompt\": \"  'flowtype' | \\\"[]\\\"\",\n",
      "    \"predicted\": \"#meta+type+name+impl+list+shape\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#financial+contribution+type\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'id' | \\\"['1092', '946']\\\"\",\n",
      "    \"predicted\": \"#activity+appeal+id+fts_internal(\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#activity+appeal+id+fts_internal\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'description' | \\\"[]\\\"\",\n",
      "    \"predicted\": \"#description+notes+v_description+list+en\",\n",
      "    \"predicted_log_prob_cutoff\": \"#description+notes\",\n",
      "    \"expected\": \"#description\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['lao', 'lao', 'lao', 'lao', 'lao', 'lao', 'lao', 'lao']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+v_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['niu', 'niu', 'niu', 'niu', 'niu', 'niu', 'niu', 'niu']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+Countries+\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'clustercode' | \\\"['2', '3', '6', '7', '9', '10', '11', nan]\\\"\",\n",
      "    \"predicted\": \"#sector+cluster+code+v_iso2\",\n",
      "    \"predicted_log_prob_cutoff\": \"#sector\",\n",
      "    \"expected\": \"#sector+cluster+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'type' | \\\"[]\\\"\",\n",
      "    \"predicted\": \"#loc+airport+type+name+i_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#loc+airport+type\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'refcode' | \\\"['[oct 10835];[undac]', 'goal hpp 2022', 'trocaire hpp 2022', 'oxfam hpp 2022', 'caid hpp 2022', 'wv hpp 2022', 'plan hpp 2022', 'con hpp 2022']\\\"\",\n",
      "    \"predicted\": \"#activity+code+refcode+v_espan\",\n",
      "    \"predicted_log_prob_cutoff\": \"#activity\",\n",
      "    \"expected\": \"#activity+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'flowtype' | \\\"['parked', 'standard', 'parked', 'standard', 'standard', 'parked', 'standard', 'parked']\\\"\",\n",
      "    \"predicted\": \"#meta+type+name+list+map+fl\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#financial+contribution+type\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'code' | \\\"['hcod23', 'hcod23', 'hcod23', 'hcod22', 'hcod22', 'hcod22', 'hcod22', 'hcod22']\\\"\",\n",
      "    \"predicted\": \"#activity+appeal+id+external+url_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#activity+appeal+id+external\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['npl', 'npl', 'npl', 'npl', 'npl', 'npl']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'source_headline' | \\\"[nan]\\\"\",\n",
      "    \"predicted\": \"#description+source+heading+invention+name+\",\n",
      "    \"predicted_log_prob_cutoff\": \"#description+source\",\n",
      "    \"expected\": \"#description\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['rou', 'rou', 'rou', 'rou', 'rou', 'rou', 'rou', 'rou']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['lux', 'lux', 'lux', 'lux', 'lux', 'lux', 'lux', 'lux']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+retain\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"[]\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country_code' | \\\"['saint pierre and miquelon', 'saint pierre and miquelon', 'saint pierre and miquelon']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+iso2\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code+v_iso2\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'code' | \\\"[]\\\"\",\n",
      "    \"predicted\": \"#activity+appeal+id+external+mis+\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#meta+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['niu', 'niu', 'niu', 'niu', 'niu', 'niu']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+p_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['npl', 'npl', 'npl', 'npl', 'npl', 'npl', 'npl', 'npl']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+num_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['npl', 'npl', 'npl', 'npl', 'npl', 'npl', 'npl', 'npl']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+hr_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['lbn', 'lbn', 'lbn', 'lbn', 'lbn', 'lbn', 'lbn', 'lbn']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2_iso2\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['slv', 'slv', 'slv', 'slv', 'slv', 'slv', 'slv', 'slv']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+folare\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'keywords' | \\\"[]\\\"\",\n",
      "    \"predicted\": \"#meta+keywords+v_desc+v_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#description+keywords\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['ukr', 'ukr', 'ukr', 'ukr', 'ukr', 'ukr', 'ukr', 'ukr']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+iso2\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['png', 'png', 'png', 'png', 'png', 'png', 'png', 'png']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'hazard type' | \\\"['flood', 'wet mass movement', 'storm', 'flood', 'flood', 'flood', 'storm', 'storm']\\\"\",\n",
      "    \"predicted\": \"#crisis+type+name+impl+list+\",\n",
      "    \"predicted_log_prob_cutoff\": \"#crisis\",\n",
      "    \"expected\": \"#crisis+type\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['tls', 'tls', 'tls', 'tls', 'tls', 'tls', 'tls', 'tls']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['bfa', 'bfa', 'bfa', 'bfa', 'bfa', 'bfa', 'bfa', 'bfa']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+enum\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'onboundary' | \\\"['shared', 'shared', 'single', 'shared', 'shared', 'shared', 'shared', 'shared']\\\"\",\n",
      "    \"predicted\": \"#affected+onbound+local+name_bubble\",\n",
      "    \"predicted_log_prob_cutoff\": \"#\",\n",
      "    \"expected\": \"#financial+direction+type\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['jor']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+v_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['gnq', 'gnq', 'gnq', 'gnq', 'gnq', 'gnq', 'gnq', 'gnq']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+dummy\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['bfa', 'bfa', 'bfa', 'bfa', 'bfa', 'bfa', 'bfa', 'bfa']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso2+exped\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['hnd', 'hnd', 'hnd', 'hnd', 'hnd', 'hnd', 'hnd', 'hnd']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+border_\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['hnd', 'hnd', 'hnd', 'hnd', 'hnd', 'hnd', 'hnd']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n",
      "{\n",
      "    \"prompt\": \"  'country (code)' | \\\"['ecu', 'ecu', 'ecu', 'ecu', 'ecu', 'ecu', 'ecu', 'ecu']\\\"\",\n",
      "    \"predicted\": \"#country+code+v_iso3+iso3\",\n",
      "    \"predicted_log_prob_cutoff\": \"#country+code+v_iso\",\n",
      "    \"expected\": \"#country+code\"\n",
      "}\n"
     ]
    }
   ],
   "source": [
    "passes = 0\n",
    "fails = 0\n",
    "for r in results:\n",
    "    if r[\"predicted_log_prob_cutoff\"] != r[\"expected\"]:\n",
    "        print(json.dumps(r, indent=4, sort_keys=False))"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Interestingly, comparing `expected` to `predicted` we see that GPT-3 actually predicts the tag and attributes most of the time, but has some extra tokens after those. So if we compare just the tag (first part with `#` prefix) it's spot-on, but when including attributes it's terrbile.\n",
    "\n",
    "I played around with model parameters to limit these extra tokens without much success. Though it was possible to get the token count exactly right for some tag types, for others that would fail. The challenge is that HXL tags can have varying numbers of attributess.\n",
    "\n",
    "### GPT-3 predictions with a post-processor to chop off unwaned extra trailing tokens\n",
    "\n",
    "For now, I as you can see above, created a little post-process which chops off the extra predicted tokens based on certain tag types. Specifically ...\n",
    "\n",
    "1. Removes duplicate attributes from GPT-3 completion\n",
    "2. Removes any tags/attributes which aren't in training set tag/attribute list\n",
    "3. For 'value' and 'airport' tags, number of attributes are limited to 3, else 2\n",
    "4. If the completion starts with `'#date+year+start','#date+year+end','#geo+precision+code'` set the HXL tag to those\n",
    "\n",
    "Not pretty, I know, and there is more work to do to control this via GPT-3, but let's see how well this works ..."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "GPT-3 results for predicted_post_processed, 10 predictions ...\n",
      "\n",
      "Just HXL tags ...\n",
      "\n",
      "Accuracy: 1.0\n",
      "Precision: 1.0\n",
      "Recall: 1.0\n",
      "F1: 1.0\n",
      "\n",
      "Tags and attributes with predicted_post_processed ...\n",
      "\n",
      "Accuracy: 0.6\n",
      "Precision: 0.6\n",
      "Recall: 0.6\n",
      "F1: 0.6\n"
     ]
    }
   ],
   "source": [
    "output_prediction_metrics(results, prediction_field=\"predicted_post_processed\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Not too bad. Predicting just the HXL tag yielded 0.97 accuracy with no discarded rows due to confidence filtering. Accuracy for tags and attributes is pretty good at 0.8, but like could be improved in future work. It does show however that a simply GPT-3 fine tuning can yield some great results."
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "## Future work\n",
    "\n",
    "This was a quick analysis, future work might include:\n",
    "\n",
    "1. Trying other models (I used 'ada' above) to see if this improves the extra tokens issue\n",
    "2. Model hyperparameter tunning. There are only 2-3, which I manually adjusted, but more work here might yield better results\n",
    "3. Perhaps try different prompt forms. For example more sampling of the data and adding in list of surounding columns for better context\n",
    "\n",
    "What is interesting is that GPT-3 would seem to have great potential in the field of meta data discovery."
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "dk_analysis",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "bb07ad760a777b0820a256b02271b7ddc47c45fb246905d25ef750a6b44c30e7"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}