python_final_1_of_2_scikit_learn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "f69689d1",
   "metadata": {},
   "source": [
    "# Data Science 101\n",
    "## [Classifier Models](https://scikit-learn.org/stable/modules/tree.html#classification)\n",
    "\n",
    "We'll use [scikit-learn](https://scikit-learn.org/stable/) to create a model that classifies some data (just created randomly) into 3 classes. This will be a good mockup of our cGMP datasets for the biosciences.\n",
    "\n",
    "A classifier model separates 2 or more classes. For example, the blue and red lines below separate the filled circles from the empty circles. Each line would be a unique classifier (some are better than others as you can see).\n",
    "\n",
    "![classifier image](https://upload.wikimedia.org/wikipedia/commons/thumb/2/20/Svm_separating_hyperplanes.png/1024px-Svm_separating_hyperplanes.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ea911e26",
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import pandas as pd\n",
    "\n",
    "import csv"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "38e050ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "import sklearn   # %pip install scikit-learn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "37962c00",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.datasets import make_classification"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "18248b2f",
   "metadata": {},
   "source": [
    "## Create some data for a classification model\n",
    "\n",
    "`scikit-learn` has a nice helper function to generate classification datasets. We'll use this to create our toy dataset. There will be 3 classes and 8 features (variables)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "fb011afd",
   "metadata": {},
   "outputs": [],
   "source": [
    "np.random.seed(816)  # Make sure we get the same random sequence whenever we run it"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "2549a5e9",
   "metadata": {},
   "outputs": [],
   "source": [
    "X, Y = make_classification(n_samples=605923, n_features=8, n_classes=3, n_informative=5)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "52656a1e",
   "metadata": {},
   "source": [
    "## Put the data into a DataFrame\n",
    "\n",
    "DataFrames make it easy to organize and massage our datasets. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "76a67624",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_names = np.array(\n",
    "    [\n",
    "        \"pH\",\n",
    "        \"concentration (μg/ml)\",\n",
    "        \"molarity (mol/L)\",\n",
    "        \"temperature (°C)\",\n",
    "        \"lipophilicity\",\n",
    "        \"volume (ml)\",\n",
    "        \"turbidity (NTU)\",\n",
    "        \"elapsed time (secs)\",\n",
    "    ]\n",
    ")\n",
    "\n",
    "df = pd.DataFrame(X, columns=feature_names)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "03022032",
   "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>pH</th>\n",
       "      <th>concentration (μg/ml)</th>\n",
       "      <th>molarity (mol/L)</th>\n",
       "      <th>temperature (°C)</th>\n",
       "      <th>lipophilicity</th>\n",
       "      <th>volume (ml)</th>\n",
       "      <th>turbidity (NTU)</th>\n",
       "      <th>elapsed time (secs)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.725232</td>\n",
       "      <td>1.945323</td>\n",
       "      <td>-1.430745</td>\n",
       "      <td>-1.478135</td>\n",
       "      <td>-1.881911</td>\n",
       "      <td>2.051609</td>\n",
       "      <td>-0.437596</td>\n",
       "      <td>-1.738268</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.177159</td>\n",
       "      <td>1.386740</td>\n",
       "      <td>-1.742781</td>\n",
       "      <td>1.535979</td>\n",
       "      <td>-2.805019</td>\n",
       "      <td>-4.131702</td>\n",
       "      <td>0.595392</td>\n",
       "      <td>0.132199</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.390952</td>\n",
       "      <td>-0.892718</td>\n",
       "      <td>-0.377981</td>\n",
       "      <td>0.032222</td>\n",
       "      <td>1.035891</td>\n",
       "      <td>0.472194</td>\n",
       "      <td>0.548471</td>\n",
       "      <td>0.350795</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-0.235868</td>\n",
       "      <td>1.776299</td>\n",
       "      <td>-0.150960</td>\n",
       "      <td>-0.686130</td>\n",
       "      <td>-1.907406</td>\n",
       "      <td>-2.756324</td>\n",
       "      <td>-0.596668</td>\n",
       "      <td>-1.010278</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.657548</td>\n",
       "      <td>0.003453</td>\n",
       "      <td>1.185180</td>\n",
       "      <td>-0.189665</td>\n",
       "      <td>-0.302958</td>\n",
       "      <td>-1.250753</td>\n",
       "      <td>-0.042773</td>\n",
       "      <td>-0.540640</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         pH  concentration (μg/ml)  molarity (mol/L)  temperature (°C)  \\\n",
       "0  0.725232               1.945323         -1.430745         -1.478135   \n",
       "1  1.177159               1.386740         -1.742781          1.535979   \n",
       "2  0.390952              -0.892718         -0.377981          0.032222   \n",
       "3 -0.235868               1.776299         -0.150960         -0.686130   \n",
       "4  0.657548               0.003453          1.185180         -0.189665   \n",
       "\n",
       "   lipophilicity  volume (ml)  turbidity (NTU)  elapsed time (secs)  \n",
       "0      -1.881911     2.051609        -0.437596            -1.738268  \n",
       "1      -2.805019    -4.131702         0.595392             0.132199  \n",
       "2       1.035891     0.472194         0.548471             0.350795  \n",
       "3      -1.907406    -2.756324        -0.596668            -1.010278  \n",
       "4      -0.302958    -1.250753        -0.042773            -0.540640  "
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(5)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "724b838d",
   "metadata": {},
   "source": [
    "### Scale and shift to make the data seem realistic\n",
    "\n",
    "I'm just randomly chosing values to scale and shift the data so that the values \"make sense\" for the parameter they represent. For example, concentrations and elapsed time should be positive numbers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "3f347eca",
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add the minimum value to everything but the 3rd column (temperature)\n",
    "# This makes those values all positive numbers.\n",
    "for column in feature_names[np.arange(len(feature_names)) != 3]:\n",
    "    df[column] += abs(df[column].min())\n",
    "\n",
    "# It's random data so let's just get it\n",
    "#       looking correctly for the feature\n",
    "df[\"temperature (°C)\"] *= 10\n",
    "df[\"temperature (°C)\"] += 5.4\n",
    "df[\"volume (ml)\"] *= 1e4\n",
    "df[\"concentration (μg/ml)\"] /= 1e4\n",
    "df[\"elapsed time (secs)\"] += 227\n",
    "df[\"elapsed time (secs)\"] *= 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "31ba2980",
   "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>pH</th>\n",
       "      <th>concentration (μg/ml)</th>\n",
       "      <th>molarity (mol/L)</th>\n",
       "      <th>temperature (°C)</th>\n",
       "      <th>lipophilicity</th>\n",
       "      <th>volume (ml)</th>\n",
       "      <th>turbidity (NTU)</th>\n",
       "      <th>elapsed time (secs)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.744047</td>\n",
       "      <td>0.000787</td>\n",
       "      <td>5.234648</td>\n",
       "      <td>-9.381354</td>\n",
       "      <td>12.528551</td>\n",
       "      <td>93655.444937</td>\n",
       "      <td>7.631413</td>\n",
       "      <td>2345.429022</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6.195974</td>\n",
       "      <td>0.000732</td>\n",
       "      <td>4.922611</td>\n",
       "      <td>20.759788</td>\n",
       "      <td>11.605443</td>\n",
       "      <td>31822.339160</td>\n",
       "      <td>8.664401</td>\n",
       "      <td>2364.133690</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.409767</td>\n",
       "      <td>0.000504</td>\n",
       "      <td>6.287411</td>\n",
       "      <td>5.722223</td>\n",
       "      <td>15.446353</td>\n",
       "      <td>77861.300135</td>\n",
       "      <td>8.617480</td>\n",
       "      <td>2366.319655</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.782947</td>\n",
       "      <td>0.000771</td>\n",
       "      <td>6.514433</td>\n",
       "      <td>-1.461302</td>\n",
       "      <td>12.503056</td>\n",
       "      <td>45576.113138</td>\n",
       "      <td>7.472341</td>\n",
       "      <td>2352.708926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.676364</td>\n",
       "      <td>0.000593</td>\n",
       "      <td>7.850572</td>\n",
       "      <td>3.503352</td>\n",
       "      <td>14.107504</td>\n",
       "      <td>60631.825191</td>\n",
       "      <td>8.026235</td>\n",
       "      <td>2357.405303</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605918</th>\n",
       "      <td>6.331137</td>\n",
       "      <td>0.000859</td>\n",
       "      <td>6.351610</td>\n",
       "      <td>-8.360471</td>\n",
       "      <td>8.022953</td>\n",
       "      <td>81362.104250</td>\n",
       "      <td>9.834977</td>\n",
       "      <td>2305.514459</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605919</th>\n",
       "      <td>5.127211</td>\n",
       "      <td>0.000445</td>\n",
       "      <td>7.143865</td>\n",
       "      <td>-0.092239</td>\n",
       "      <td>15.190061</td>\n",
       "      <td>57069.762130</td>\n",
       "      <td>9.665127</td>\n",
       "      <td>2353.064857</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605920</th>\n",
       "      <td>6.676017</td>\n",
       "      <td>0.000845</td>\n",
       "      <td>6.818992</td>\n",
       "      <td>-2.759908</td>\n",
       "      <td>12.436351</td>\n",
       "      <td>67069.699844</td>\n",
       "      <td>6.266615</td>\n",
       "      <td>2355.959926</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605921</th>\n",
       "      <td>3.071569</td>\n",
       "      <td>0.000594</td>\n",
       "      <td>6.693396</td>\n",
       "      <td>13.432805</td>\n",
       "      <td>14.681529</td>\n",
       "      <td>84802.051315</td>\n",
       "      <td>7.508070</td>\n",
       "      <td>2373.199376</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605922</th>\n",
       "      <td>5.263818</td>\n",
       "      <td>0.000395</td>\n",
       "      <td>3.902334</td>\n",
       "      <td>-3.057376</td>\n",
       "      <td>15.414881</td>\n",
       "      <td>96116.563337</td>\n",
       "      <td>11.204406</td>\n",
       "      <td>2348.976253</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>605923 rows × 8 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              pH  concentration (μg/ml)  molarity (mol/L)  temperature (°C)  \\\n",
       "0       5.744047               0.000787          5.234648         -9.381354   \n",
       "1       6.195974               0.000732          4.922611         20.759788   \n",
       "2       5.409767               0.000504          6.287411          5.722223   \n",
       "3       4.782947               0.000771          6.514433         -1.461302   \n",
       "4       5.676364               0.000593          7.850572          3.503352   \n",
       "...          ...                    ...               ...               ...   \n",
       "605918  6.331137               0.000859          6.351610         -8.360471   \n",
       "605919  5.127211               0.000445          7.143865         -0.092239   \n",
       "605920  6.676017               0.000845          6.818992         -2.759908   \n",
       "605921  3.071569               0.000594          6.693396         13.432805   \n",
       "605922  5.263818               0.000395          3.902334         -3.057376   \n",
       "\n",
       "        lipophilicity   volume (ml)  turbidity (NTU)  elapsed time (secs)  \n",
       "0           12.528551  93655.444937         7.631413          2345.429022  \n",
       "1           11.605443  31822.339160         8.664401          2364.133690  \n",
       "2           15.446353  77861.300135         8.617480          2366.319655  \n",
       "3           12.503056  45576.113138         7.472341          2352.708926  \n",
       "4           14.107504  60631.825191         8.026235          2357.405303  \n",
       "...               ...           ...              ...                  ...  \n",
       "605918       8.022953  81362.104250         9.834977          2305.514459  \n",
       "605919      15.190061  57069.762130         9.665127          2353.064857  \n",
       "605920      12.436351  67069.699844         6.266615          2355.959926  \n",
       "605921      14.681529  84802.051315         7.508070          2373.199376  \n",
       "605922      15.414881  96116.563337        11.204406          2348.976253  \n",
       "\n",
       "[605923 rows x 8 columns]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "79b07907",
   "metadata": {},
   "source": [
    "### Pandas makes plotting lots of columns easy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "221a02d6",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 2000x1500 with 9 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "df.hist(figsize=(20,15));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "d48353d8",
   "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>pH</th>\n",
       "      <th>concentration (μg/ml)</th>\n",
       "      <th>molarity (mol/L)</th>\n",
       "      <th>temperature (°C)</th>\n",
       "      <th>lipophilicity</th>\n",
       "      <th>volume (ml)</th>\n",
       "      <th>turbidity (NTU)</th>\n",
       "      <th>elapsed time (secs)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>count</th>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "      <td>605923.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>mean</th>\n",
       "      <td>5.017745</td>\n",
       "      <td>0.000626</td>\n",
       "      <td>6.997431</td>\n",
       "      <td>2.089461</td>\n",
       "      <td>13.850906</td>\n",
       "      <td>76470.110018</td>\n",
       "      <td>8.006114</td>\n",
       "      <td>2356.158802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>std</th>\n",
       "      <td>1.001131</td>\n",
       "      <td>0.000158</td>\n",
       "      <td>1.665213</td>\n",
       "      <td>16.492694</td>\n",
       "      <td>2.264557</td>\n",
       "      <td>17076.304671</td>\n",
       "      <td>1.781705</td>\n",
       "      <td>15.200348</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>min</th>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>-84.096259</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>2270.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25%</th>\n",
       "      <td>4.342642</td>\n",
       "      <td>0.000516</td>\n",
       "      <td>5.797549</td>\n",
       "      <td>-9.284253</td>\n",
       "      <td>12.506152</td>\n",
       "      <td>64699.857456</td>\n",
       "      <td>6.768615</td>\n",
       "      <td>2346.132154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>50%</th>\n",
       "      <td>5.017004</td>\n",
       "      <td>0.000631</td>\n",
       "      <td>6.965851</td>\n",
       "      <td>2.122003</td>\n",
       "      <td>13.942072</td>\n",
       "      <td>76797.404903</td>\n",
       "      <td>8.064196</td>\n",
       "      <td>2355.513485</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>75%</th>\n",
       "      <td>5.692343</td>\n",
       "      <td>0.000737</td>\n",
       "      <td>8.143590</td>\n",
       "      <td>13.750541</td>\n",
       "      <td>15.292081</td>\n",
       "      <td>88355.852928</td>\n",
       "      <td>9.289894</td>\n",
       "      <td>2365.817968</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>max</th>\n",
       "      <td>9.513773</td>\n",
       "      <td>0.001448</td>\n",
       "      <td>15.863927</td>\n",
       "      <td>71.341922</td>\n",
       "      <td>27.976120</td>\n",
       "      <td>156922.597691</td>\n",
       "      <td>16.366807</td>\n",
       "      <td>2440.762676</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                  pH  concentration (μg/ml)  molarity (mol/L)  \\\n",
       "count  605923.000000          605923.000000     605923.000000   \n",
       "mean        5.017745               0.000626          6.997431   \n",
       "std         1.001131               0.000158          1.665213   \n",
       "min         0.000000               0.000000          0.000000   \n",
       "25%         4.342642               0.000516          5.797549   \n",
       "50%         5.017004               0.000631          6.965851   \n",
       "75%         5.692343               0.000737          8.143590   \n",
       "max         9.513773               0.001448         15.863927   \n",
       "\n",
       "       temperature (°C)  lipophilicity    volume (ml)  turbidity (NTU)  \\\n",
       "count     605923.000000  605923.000000  605923.000000    605923.000000   \n",
       "mean           2.089461      13.850906   76470.110018         8.006114   \n",
       "std           16.492694       2.264557   17076.304671         1.781705   \n",
       "min          -84.096259       0.000000       0.000000         0.000000   \n",
       "25%           -9.284253      12.506152   64699.857456         6.768615   \n",
       "50%            2.122003      13.942072   76797.404903         8.064196   \n",
       "75%           13.750541      15.292081   88355.852928         9.289894   \n",
       "max           71.341922      27.976120  156922.597691        16.366807   \n",
       "\n",
       "       elapsed time (secs)  \n",
       "count        605923.000000  \n",
       "mean           2356.158802  \n",
       "std              15.200348  \n",
       "min            2270.000000  \n",
       "25%            2346.132154  \n",
       "50%            2355.513485  \n",
       "75%            2365.817968  \n",
       "max            2440.762676  "
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.describe()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "cc36db85",
   "metadata": {},
   "source": [
    "## Add the classification label to the DataFrame\n",
    "\n",
    "We'll call this `Antibody affinity` to make it seem realistic."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "ae5366f4",
   "metadata": {},
   "outputs": [],
   "source": [
    "def class_name(value):\n",
    "    \"\"\"Assign a class name based on the value\n",
    "    \n",
    "    We've assume 3 classes (low, medium, and high)\n",
    "    \"\"\"\n",
    "\n",
    "    if value == 0:\n",
    "        return \"Low\"\n",
    "    elif value == 1:\n",
    "        return \"Medium\"\n",
    "    else:\n",
    "        return \"High\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "9de41fed",
   "metadata": {},
   "outputs": [],
   "source": [
    "prediction_name = \"Antibody affinity\"\n",
    "df[prediction_name] = list(map(class_name, Y))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "1c61a07c",
   "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>pH</th>\n",
       "      <th>concentration (μg/ml)</th>\n",
       "      <th>molarity (mol/L)</th>\n",
       "      <th>temperature (°C)</th>\n",
       "      <th>lipophilicity</th>\n",
       "      <th>volume (ml)</th>\n",
       "      <th>turbidity (NTU)</th>\n",
       "      <th>elapsed time (secs)</th>\n",
       "      <th>Antibody affinity</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>5.744047</td>\n",
       "      <td>0.000787</td>\n",
       "      <td>5.234648</td>\n",
       "      <td>-9.381354</td>\n",
       "      <td>12.528551</td>\n",
       "      <td>93655.444937</td>\n",
       "      <td>7.631413</td>\n",
       "      <td>2345.429022</td>\n",
       "      <td>Low</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>6.195974</td>\n",
       "      <td>0.000732</td>\n",
       "      <td>4.922611</td>\n",
       "      <td>20.759788</td>\n",
       "      <td>11.605443</td>\n",
       "      <td>31822.339160</td>\n",
       "      <td>8.664401</td>\n",
       "      <td>2364.133690</td>\n",
       "      <td>High</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>5.409767</td>\n",
       "      <td>0.000504</td>\n",
       "      <td>6.287411</td>\n",
       "      <td>5.722223</td>\n",
       "      <td>15.446353</td>\n",
       "      <td>77861.300135</td>\n",
       "      <td>8.617480</td>\n",
       "      <td>2366.319655</td>\n",
       "      <td>Low</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>4.782947</td>\n",
       "      <td>0.000771</td>\n",
       "      <td>6.514433</td>\n",
       "      <td>-1.461302</td>\n",
       "      <td>12.503056</td>\n",
       "      <td>45576.113138</td>\n",
       "      <td>7.472341</td>\n",
       "      <td>2352.708926</td>\n",
       "      <td>Medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>5.676364</td>\n",
       "      <td>0.000593</td>\n",
       "      <td>7.850572</td>\n",
       "      <td>3.503352</td>\n",
       "      <td>14.107504</td>\n",
       "      <td>60631.825191</td>\n",
       "      <td>8.026235</td>\n",
       "      <td>2357.405303</td>\n",
       "      <td>Medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605918</th>\n",
       "      <td>6.331137</td>\n",
       "      <td>0.000859</td>\n",
       "      <td>6.351610</td>\n",
       "      <td>-8.360471</td>\n",
       "      <td>8.022953</td>\n",
       "      <td>81362.104250</td>\n",
       "      <td>9.834977</td>\n",
       "      <td>2305.514459</td>\n",
       "      <td>Medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605919</th>\n",
       "      <td>5.127211</td>\n",
       "      <td>0.000445</td>\n",
       "      <td>7.143865</td>\n",
       "      <td>-0.092239</td>\n",
       "      <td>15.190061</td>\n",
       "      <td>57069.762130</td>\n",
       "      <td>9.665127</td>\n",
       "      <td>2353.064857</td>\n",
       "      <td>Medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605920</th>\n",
       "      <td>6.676017</td>\n",
       "      <td>0.000845</td>\n",
       "      <td>6.818992</td>\n",
       "      <td>-2.759908</td>\n",
       "      <td>12.436351</td>\n",
       "      <td>67069.699844</td>\n",
       "      <td>6.266615</td>\n",
       "      <td>2355.959926</td>\n",
       "      <td>Medium</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605921</th>\n",
       "      <td>3.071569</td>\n",
       "      <td>0.000594</td>\n",
       "      <td>6.693396</td>\n",
       "      <td>13.432805</td>\n",
       "      <td>14.681529</td>\n",
       "      <td>84802.051315</td>\n",
       "      <td>7.508070</td>\n",
       "      <td>2373.199376</td>\n",
       "      <td>Low</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>605922</th>\n",
       "      <td>5.263818</td>\n",
       "      <td>0.000395</td>\n",
       "      <td>3.902334</td>\n",
       "      <td>-3.057376</td>\n",
       "      <td>15.414881</td>\n",
       "      <td>96116.563337</td>\n",
       "      <td>11.204406</td>\n",
       "      <td>2348.976253</td>\n",
       "      <td>Low</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>605923 rows × 9 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "              pH  concentration (μg/ml)  molarity (mol/L)  temperature (°C)  \\\n",
       "0       5.744047               0.000787          5.234648         -9.381354   \n",
       "1       6.195974               0.000732          4.922611         20.759788   \n",
       "2       5.409767               0.000504          6.287411          5.722223   \n",
       "3       4.782947               0.000771          6.514433         -1.461302   \n",
       "4       5.676364               0.000593          7.850572          3.503352   \n",
       "...          ...                    ...               ...               ...   \n",
       "605918  6.331137               0.000859          6.351610         -8.360471   \n",
       "605919  5.127211               0.000445          7.143865         -0.092239   \n",
       "605920  6.676017               0.000845          6.818992         -2.759908   \n",
       "605921  3.071569               0.000594          6.693396         13.432805   \n",
       "605922  5.263818               0.000395          3.902334         -3.057376   \n",
       "\n",
       "        lipophilicity   volume (ml)  turbidity (NTU)  elapsed time (secs)  \\\n",
       "0           12.528551  93655.444937         7.631413          2345.429022   \n",
       "1           11.605443  31822.339160         8.664401          2364.133690   \n",
       "2           15.446353  77861.300135         8.617480          2366.319655   \n",
       "3           12.503056  45576.113138         7.472341          2352.708926   \n",
       "4           14.107504  60631.825191         8.026235          2357.405303   \n",
       "...               ...           ...              ...                  ...   \n",
       "605918       8.022953  81362.104250         9.834977          2305.514459   \n",
       "605919      15.190061  57069.762130         9.665127          2353.064857   \n",
       "605920      12.436351  67069.699844         6.266615          2355.959926   \n",
       "605921      14.681529  84802.051315         7.508070          2373.199376   \n",
       "605922      15.414881  96116.563337        11.204406          2348.976253   \n",
       "\n",
       "       Antibody affinity  \n",
       "0                    Low  \n",
       "1                   High  \n",
       "2                    Low  \n",
       "3                 Medium  \n",
       "4                 Medium  \n",
       "...                  ...  \n",
       "605918            Medium  \n",
       "605919            Medium  \n",
       "605920            Medium  \n",
       "605921               Low  \n",
       "605922               Low  \n",
       "\n",
       "[605923 rows x 9 columns]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a1cf68f8",
   "metadata": {},
   "source": [
    "## Save the data to a CSV file\n",
    "\n",
    "Realistically you'll get data as a CSV. In some cases, you'll have enough memory to load in the whole CSV. Nevertheless, I'll show how to create a data loader that will scale with whatever size dataset you use (and can be parallelized across compute)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "b6618649",
   "metadata": {},
   "outputs": [],
   "source": [
    "csv_filename = \"toy_cGMP_dataset.csv\" \n",
    "df.to_csv(csv_filename, index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "9851990c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "pH,concentration (μg/ml),molarity (mol/L),temperature (°C),lipophilicity,volume (ml),turbidity (NTU),elapsed time (secs),Antibody affinity\r\n",
      "5.744046917034562,0.0007874893859659413,5.234647589904735,-9.381354161554542,12.528550546588875,93655.44493728782,7.631412915799664,2345.429022336766,Low\r\n",
      "6.195974069133236,0.0007316310797757877,4.9226113990165565,20.759787752823108,11.605442862909678,31822.339159810548,8.664400898822246,2364.1336895249647,High\r\n",
      "5.409766788724181,0.0005036852507149984,6.287411437851176,5.722223439345846,15.446353001978782,77861.30013459167,8.61747954184812,2366.3196550277926,Low\r\n",
      "4.7829470760548825,0.0007705869671046501,6.514432619808776,-1.4613023978987307,12.503055902615785,45576.11313756699,7.472340890764882,2352.708925680164,Medium\r\n",
      "5.676363633556237,0.0005933024407799599,7.85057187038264,3.503351517122042,14.10750405510838,60631.825190692725,8.026235278506693,2357.405303106366,Medium\r\n",
      "4.544998786858278,0.000788394916575371,5.481094500400836,-0.3232518740276795,13.756432596856978,75680.8244571045,6.369683833549491,2367.8928694102597,Low\r\n",
      "5.939616075448547,0.000814733582089522,4.238695463345631,-22.893694959643042,12.537542981404743,84508.73683900294,8.064849605529005,2335.52970504125,Low\r\n",
      "6.365931660735418,0.00025506965375256827,5.239360063963112,-0.31268862746232884,17.42507153180413,116707.60469952963,10.938264905364626,2358.6339363457787,Low\r\n",
      "4.2916408536519155,0.0008355366483480559,5.659683329326514,-34.3703583588642,12.151614876563121,108993.6407748543,7.807293129018748,2319.644261796635,High\r\n",
      "5.110047833685457,0.00045482915795870565,8.738286940262444,-11.044930793268511,18.48034299250719,57767.68687060997,6.705757417135921,2375.774747843454,Medium\r\n",
      "6.255375702973213,0.0006433699724276163,6.566685523150818,23.285370120293912,13.743397149608523,70146.66696462227,7.356433032722675,2378.3104848155654,Low\r\n",
      "6.4886355403658555,0.0008941176822537592,5.3151589523319265,-4.931340780069444,12.014525116717623,85771.32545141233,6.38796320300999,2353.971011396756,Low\r\n",
      "4.169116090456363,0.001069513866748022,5.5181141887624765,4.3083429924285355,8.887489484952615,97022.48004700358,5.931978315445862,2345.1603282859637,High\r\n",
      "5.508270912661684,0.000402971248530904,5.717589700252887,0.9237256898221329,16.10430078903139,91111.09622313606,9.811913685680711,2359.5843874224684,Low\r\n"
     ]
    }
   ],
   "source": [
    "!head -15 $csv_filename"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "6c9e5660",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "90M\ttoy_cGMP_dataset.csv\r\n"
     ]
    }
   ],
   "source": [
    "# Size of the file\n",
    "!du -h $csv_filename"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "878a8084",
   "metadata": {},
   "source": [
    "## Let's create our own data loader\n",
    "\n",
    "Data loaders are [generators](https://wiki.python.org/moin/Generators). Note the use of the `yield` command here. It allows the program execution to return to the line that called it and proceed. When the generator is called again, then it picks up where it left off. So it is giving us just one line at a time (or we can have it give us a few lines at a time).\n",
    "\n",
    "Data loaders are great because they just load in a batch at a time. This allows us to work on massively-sized datasets. It also allows us to spread our compute over many computers if we want."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "3478d6fe",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_data(filename):\n",
    "    \"\"\"Gets data from CSV\"\"\"\n",
    "    \n",
    "    with open(filename, \"r\") as csvfile:\n",
    "        \n",
    "        reader = csv.DictReader(csvfile)\n",
    "        for row in reader:\n",
    "            yield row\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "cc6d87af",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_loader = get_data(csv_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "ff2391eb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<generator object get_data at 0x7f6d58716810>"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_loader"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ab22efe9",
   "metadata": {},
   "source": [
    "### First call gets us the first 5 lines of the CSV file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "6f118ae0",
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'pH': '5.744046917034562', 'concentration (μg/ml)': '0.0007874893859659413', 'molarity (mol/L)': '5.234647589904735', 'temperature (°C)': '-9.381354161554542', 'lipophilicity': '12.528550546588875', 'volume (ml)': '93655.44493728782', 'turbidity (NTU)': '7.631412915799664', 'elapsed time (secs)': '2345.429022336766', 'Antibody affinity': 'Low'}\n",
      "{'pH': '6.195974069133236', 'concentration (μg/ml)': '0.0007316310797757877', 'molarity (mol/L)': '4.9226113990165565', 'temperature (°C)': '20.759787752823108', 'lipophilicity': '11.605442862909678', 'volume (ml)': '31822.339159810548', 'turbidity (NTU)': '8.664400898822246', 'elapsed time (secs)': '2364.1336895249647', 'Antibody affinity': 'High'}\n",
      "{'pH': '5.409766788724181', 'concentration (μg/ml)': '0.0005036852507149984', 'molarity (mol/L)': '6.287411437851176', 'temperature (°C)': '5.722223439345846', 'lipophilicity': '15.446353001978782', 'volume (ml)': '77861.30013459167', 'turbidity (NTU)': '8.61747954184812', 'elapsed time (secs)': '2366.3196550277926', 'Antibody affinity': 'Low'}\n",
      "{'pH': '4.7829470760548825', 'concentration (μg/ml)': '0.0007705869671046501', 'molarity (mol/L)': '6.514432619808776', 'temperature (°C)': '-1.4613023978987307', 'lipophilicity': '12.503055902615785', 'volume (ml)': '45576.11313756699', 'turbidity (NTU)': '7.472340890764882', 'elapsed time (secs)': '2352.708925680164', 'Antibody affinity': 'Medium'}\n",
      "{'pH': '5.676363633556237', 'concentration (μg/ml)': '0.0005933024407799599', 'molarity (mol/L)': '7.85057187038264', 'temperature (°C)': '3.503351517122042', 'lipophilicity': '14.10750405510838', 'volume (ml)': '60631.825190692725', 'turbidity (NTU)': '8.026235278506693', 'elapsed time (secs)': '2357.405303106366', 'Antibody affinity': 'Medium'}\n"
     ]
    }
   ],
   "source": [
    "rows = []\n",
    "for idx, row in enumerate(data_loader):\n",
    "    \n",
    "    print(row)\n",
    "    rows.append(row[\"pH\"])\n",
    "    \n",
    "    if idx >= 4:\n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "1f145839",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['5.744046917034562', '6.195974069133236', '5.409766788724181', '4.7829470760548825', '5.676363633556237']\n"
     ]
    }
   ],
   "source": [
    "print(rows)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "813a3183",
   "metadata": {},
   "source": [
    "#### Sanity check\n",
    "Our values should be something like this"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "df466ca6",
   "metadata": {},
   "source": [
    "5.744046917034562,0.0007874893859659413,5.234647589904735,-9.381354161554542,12.528550546588875,93655.44493728782,7.631412915799664,2345.429022336766,Low\n",
    "6.195974069133236,0.0007316310797757877,4.9226113990165565,20.759787752823108,11.605442862909678,31822.339159810548,8.664400898822246,2364.1336895249647,High\n",
    "5.409766788724181,0.0005036852507149984,6.287411437851176,5.722223439345846,15.446353001978782,77861.30013459167,8.61747954184812,2366.3196550277926,Low\n",
    "4.7829470760548825,0.0007705869671046501,6.514432619808776,-1.4613023978987307,12.503055902615785,45576.11313756699,7.472340890764882,2352.708925680164,Medium\n",
    "5.676363633556237,0.0005933024407799599,7.85057187038264,3.503351517122042,14.10750405510838,60631.825190692725,8.026235278506693,2357.405303106366,Medium"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "29591ddb",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "5.744046917034562,0.0007874893859659413,5.234647589904735,-9.381354161554542,12.528550546588875,93655.44493728782,7.631412915799664,2345.429022336766,Low\r\n",
      "6.195974069133236,0.0007316310797757877,4.9226113990165565,20.759787752823108,11.605442862909678,31822.339159810548,8.664400898822246,2364.1336895249647,High\r\n",
      "5.409766788724181,0.0005036852507149984,6.287411437851176,5.722223439345846,15.446353001978782,77861.30013459167,8.61747954184812,2366.3196550277926,Low\r\n",
      "4.7829470760548825,0.0007705869671046501,6.514432619808776,-1.4613023978987307,12.503055902615785,45576.11313756699,7.472340890764882,2352.708925680164,Medium\r\n",
      "5.676363633556237,0.0005933024407799599,7.85057187038264,3.503351517122042,14.10750405510838,60631.825190692725,8.026235278506693,2357.405303106366,Medium\r\n"
     ]
    }
   ],
   "source": [
    "!head -6 $csv_filename | tail -5   # 6 because we skip the header line"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "49c1b68f",
   "metadata": {},
   "source": [
    "### We can continue with other code if we want"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "9ffa2c40",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3\n"
     ]
    }
   ],
   "source": [
    "a = 1\n",
    "b = 2\n",
    "c = a + b\n",
    "\n",
    "print(c)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "a250b55d",
   "metadata": {},
   "source": [
    "### Second call gets us the next 5 lines of the CSV file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "c7c09fc3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'pH': '4.544998786858278', 'concentration (μg/ml)': '0.000788394916575371', 'molarity (mol/L)': '5.481094500400836', 'temperature (°C)': '-0.3232518740276795', 'lipophilicity': '13.756432596856978', 'volume (ml)': '75680.8244571045', 'turbidity (NTU)': '6.369683833549491', 'elapsed time (secs)': '2367.8928694102597', 'Antibody affinity': 'Low'}\n",
      "{'pH': '5.939616075448547', 'concentration (μg/ml)': '0.000814733582089522', 'molarity (mol/L)': '4.238695463345631', 'temperature (°C)': '-22.893694959643042', 'lipophilicity': '12.537542981404743', 'volume (ml)': '84508.73683900294', 'turbidity (NTU)': '8.064849605529005', 'elapsed time (secs)': '2335.52970504125', 'Antibody affinity': 'Low'}\n",
      "{'pH': '6.365931660735418', 'concentration (μg/ml)': '0.00025506965375256827', 'molarity (mol/L)': '5.239360063963112', 'temperature (°C)': '-0.31268862746232884', 'lipophilicity': '17.42507153180413', 'volume (ml)': '116707.60469952963', 'turbidity (NTU)': '10.938264905364626', 'elapsed time (secs)': '2358.6339363457787', 'Antibody affinity': 'Low'}\n",
      "{'pH': '4.2916408536519155', 'concentration (μg/ml)': '0.0008355366483480559', 'molarity (mol/L)': '5.659683329326514', 'temperature (°C)': '-34.3703583588642', 'lipophilicity': '12.151614876563121', 'volume (ml)': '108993.6407748543', 'turbidity (NTU)': '7.807293129018748', 'elapsed time (secs)': '2319.644261796635', 'Antibody affinity': 'High'}\n",
      "{'pH': '5.110047833685457', 'concentration (μg/ml)': '0.00045482915795870565', 'molarity (mol/L)': '8.738286940262444', 'temperature (°C)': '-11.044930793268511', 'lipophilicity': '18.48034299250719', 'volume (ml)': '57767.68687060997', 'turbidity (NTU)': '6.705757417135921', 'elapsed time (secs)': '2375.774747843454', 'Antibody affinity': 'Medium'}\n"
     ]
    }
   ],
   "source": [
    "rows = []\n",
    "for idx, row in enumerate(data_loader):\n",
    "    \n",
    "    print(row)\n",
    "    rows.append(row[\"pH\"])\n",
    "    \n",
    "    if idx >= 4:\n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "49f816c4",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['4.544998786858278', '5.939616075448547', '6.365931660735418', '4.2916408536519155', '5.110047833685457']\n"
     ]
    }
   ],
   "source": [
    "print(rows)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f6007861",
   "metadata": {},
   "source": [
    "#### Sanity check\n",
    "Our values should be something like this"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5c834e01",
   "metadata": {},
   "source": [
    "4.544998786858278,0.000788394916575371,5.481094500400836,-0.3232518740276795,13.756432596856978,75680.8244571045,6.369683833549491,2367.8928694102597,Low\n",
    "5.939616075448547,0.000814733582089522,4.238695463345631,-22.893694959643042,12.537542981404743,84508.73683900294,8.064849605529005,2335.52970504125,Low\n",
    "6.365931660735418,0.00025506965375256827,5.239360063963112,-0.31268862746232884,17.42507153180413,116707.60469952963,10.938264905364626,2358.6339363457787,Low\n",
    "4.2916408536519155,0.0008355366483480559,5.659683329326514,-34.3703583588642,12.151614876563121,108993.6407748543,7.807293129018748,2319.644261796635,High\n",
    "5.110047833685457,0.00045482915795870565,8.738286940262444,-11.044930793268511,18.48034299250719,57767.68687060997,6.705757417135921,2375.774747843454,Medium"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "1071bb9d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4.544998786858278,0.000788394916575371,5.481094500400836,-0.3232518740276795,13.756432596856978,75680.8244571045,6.369683833549491,2367.8928694102597,Low\r\n",
      "5.939616075448547,0.000814733582089522,4.238695463345631,-22.893694959643042,12.537542981404743,84508.73683900294,8.064849605529005,2335.52970504125,Low\r\n",
      "6.365931660735418,0.00025506965375256827,5.239360063963112,-0.31268862746232884,17.42507153180413,116707.60469952963,10.938264905364626,2358.6339363457787,Low\r\n",
      "4.2916408536519155,0.0008355366483480559,5.659683329326514,-34.3703583588642,12.151614876563121,108993.6407748543,7.807293129018748,2319.644261796635,High\r\n",
      "5.110047833685457,0.00045482915795870565,8.738286940262444,-11.044930793268511,18.48034299250719,57767.68687060997,6.705757417135921,2375.774747843454,Medium\r\n"
     ]
    }
   ],
   "source": [
    "!head -11 $csv_filename | tail -5  # 11 because we skip the header line"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "7731accc",
   "metadata": {},
   "source": [
    "### Third call gets us the next 5 lines of the CSV file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "2b19992c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'pH': '6.255375702973213', 'concentration (μg/ml)': '0.0006433699724276163', 'molarity (mol/L)': '6.566685523150818', 'temperature (°C)': '23.285370120293912', 'lipophilicity': '13.743397149608523', 'volume (ml)': '70146.66696462227', 'turbidity (NTU)': '7.356433032722675', 'elapsed time (secs)': '2378.3104848155654', 'Antibody affinity': 'Low'}\n",
      "{'pH': '6.4886355403658555', 'concentration (μg/ml)': '0.0008941176822537592', 'molarity (mol/L)': '5.3151589523319265', 'temperature (°C)': '-4.931340780069444', 'lipophilicity': '12.014525116717623', 'volume (ml)': '85771.32545141233', 'turbidity (NTU)': '6.38796320300999', 'elapsed time (secs)': '2353.971011396756', 'Antibody affinity': 'Low'}\n",
      "{'pH': '4.169116090456363', 'concentration (μg/ml)': '0.001069513866748022', 'molarity (mol/L)': '5.5181141887624765', 'temperature (°C)': '4.3083429924285355', 'lipophilicity': '8.887489484952615', 'volume (ml)': '97022.48004700358', 'turbidity (NTU)': '5.931978315445862', 'elapsed time (secs)': '2345.1603282859637', 'Antibody affinity': 'High'}\n",
      "{'pH': '5.508270912661684', 'concentration (μg/ml)': '0.000402971248530904', 'molarity (mol/L)': '5.717589700252887', 'temperature (°C)': '0.9237256898221329', 'lipophilicity': '16.10430078903139', 'volume (ml)': '91111.09622313606', 'turbidity (NTU)': '9.811913685680711', 'elapsed time (secs)': '2359.5843874224684', 'Antibody affinity': 'Low'}\n",
      "{'pH': '5.754263641061878', 'concentration (μg/ml)': '0.0005844853548741785', 'molarity (mol/L)': '5.727444243408503', 'temperature (°C)': '-6.781310265361496', 'lipophilicity': '14.304511492122353', 'volume (ml)': '75674.00466460997', 'turbidity (NTU)': '8.975531032483827', 'elapsed time (secs)': '2349.4343846434704', 'Antibody affinity': 'Low'}\n"
     ]
    }
   ],
   "source": [
    "rows = []\n",
    "for idx, row in enumerate(data_loader):\n",
    "    \n",
    "    print(row)\n",
    "    rows.append(row[\"pH\"])\n",
    "    \n",
    "    if idx >= 4:\n",
    "        break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "e9acaf55",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['6.255375702973213', '6.4886355403658555', '4.169116090456363', '5.508270912661684', '5.754263641061878']\n"
     ]
    }
   ],
   "source": [
    "print(rows)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "aacdc699",
   "metadata": {},
   "source": [
    "#### Sanity check\n",
    "Our values should be something like this"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "5b64207d",
   "metadata": {},
   "source": [
    "6.255375702973213,0.0006433699724276163,6.566685523150818,23.285370120293912,13.743397149608523,70146.66696462227,7.356433032722675,2378.3104848155654,Low\n",
    "6.4886355403658555,0.0008941176822537592,5.3151589523319265,-4.931340780069444,12.014525116717623,85771.32545141233,6.38796320300999,2353.971011396756,Low\n",
    "4.169116090456363,0.001069513866748022,5.5181141887624765,4.3083429924285355,8.887489484952615,97022.48004700358,5.931978315445862,2345.1603282859637,High\n",
    "5.508270912661684,0.000402971248530904,5.717589700252887,0.9237256898221329,16.10430078903139,91111.09622313606,9.811913685680711,2359.5843874224684,Low\n",
    "5.754263641061878,0.0005844853548741785,5.727444243408503,-6.781310265361496,14.304511492122353,75674.00466460997,8.975531032483827,2349.4343846434704,Low"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "79569063",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "6.255375702973213,0.0006433699724276163,6.566685523150818,23.285370120293912,13.743397149608523,70146.66696462227,7.356433032722675,2378.3104848155654,Low\r\n",
      "6.4886355403658555,0.0008941176822537592,5.3151589523319265,-4.931340780069444,12.014525116717623,85771.32545141233,6.38796320300999,2353.971011396756,Low\r\n",
      "4.169116090456363,0.001069513866748022,5.5181141887624765,4.3083429924285355,8.887489484952615,97022.48004700358,5.931978315445862,2345.1603282859637,High\r\n",
      "5.508270912661684,0.000402971248530904,5.717589700252887,0.9237256898221329,16.10430078903139,91111.09622313606,9.811913685680711,2359.5843874224684,Low\r\n",
      "5.754263641061878,0.0005844853548741785,5.727444243408503,-6.781310265361496,14.304511492122353,75674.00466460997,8.975531032483827,2349.4343846434704,Low\r\n"
     ]
    }
   ],
   "source": [
    "!head -16 $csv_filename | tail -5  # 16 because we skip the header line"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbd7cb22",
   "metadata": {},
   "source": [
    "## Putting it altogether\n",
    "\n",
    "Let's get a data loader generator that grabs one batch at a time.\n",
    "\n",
    "Note that although this data loader is specific to loading from a CSV we could easily add code to use the CSV data to load other files and then serve those to the generator. For example, our CSV could be a list of filenames and classes. \n",
    "\n",
    "```\n",
    "filename,molecule class\n",
    "path1/path2/image123.tif,dna\n",
    "path3/image234.jpg,rna\n",
    "image789.jpg,protein\n",
    "...\n",
    "path7/path8/path9/image999.jpg,antibody\n",
    "```\n",
    "\n",
    "We would just construct the `x_values` in the `yield` statement to be loaded images using the filenames from the CSV.\n",
    "\n",
    "`x_values.append(load_image(row[\"filename\"]))` where `load_image()` is our code for loading images based on the filename."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "81374c08",
   "metadata": {},
   "outputs": [],
   "source": [
    "feature_names = np.array(\n",
    "    [\n",
    "        \"pH\",\n",
    "        \"concentration (μg/ml)\",\n",
    "        \"molarity (mol/L)\",\n",
    "        \"temperature (°C)\",\n",
    "        \"lipophilicity\",\n",
    "        \"volume (ml)\",\n",
    "        \"turbidity (NTU)\",\n",
    "        \"elapsed time (secs)\",\n",
    "    ]\n",
    ")\n",
    "\n",
    "prediction_name = \"Antibody affinity\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "ce9ecd25",
   "metadata": {},
   "outputs": [],
   "source": [
    "class my_great_data_loader(object):\n",
    "    \"\"\"A data loader for our CSV files\"\"\"\n",
    "\n",
    "    def __init__(\n",
    "        self,\n",
    "        filename: str,\n",
    "        batch_size: int = 32,\n",
    "        feature_names=feature_names,\n",
    "        prediction_name: str = prediction_name,\n",
    "        rows_to_use=None,\n",
    "    ):\n",
    "        \"\"\"Initializer. This is called once at the instantiation of the object\n",
    "\n",
    "        Args:\n",
    "            filename(str): Name of the csv file\n",
    "            batch_size(int): Size of one batch (default=32 samples)\n",
    "            feature_names(List[str]): List of strings\n",
    "            prediction_name(str): The name of the column to predict\n",
    "            rows_to_use: List of indexes to use (for train/test/validation)\n",
    "        \"\"\"\n",
    "\n",
    "        self.filename = filename\n",
    "        self.batch_size = batch_size\n",
    "        self.total_rows = self.calculate_total_number_of_rows()\n",
    "        self.total_num_batches = self.calculate_total_number_of_batches()\n",
    "        self.feature_names = feature_names\n",
    "        self.prediction_name = prediction_name\n",
    "\n",
    "        # For train/test/validation splits we may want\n",
    "        #    to specify which rows are used\n",
    "        if rows_to_use is None:\n",
    "            self.rows_to_use = np.arange(self.total_rows)\n",
    "        else:\n",
    "            self.rows_to_use = rows_to_use\n",
    "\n",
    "        self.current_row = 0\n",
    "\n",
    "        self.reader = self.set_reader()\n",
    "\n",
    "    def __iter__(self):\n",
    "        return self\n",
    "\n",
    "    def __next__(self):\n",
    "        \"\"\"Get next batch\"\"\"\n",
    "\n",
    "        return self.get_next_batch()\n",
    "\n",
    "    def __len__(self):\n",
    "        \"\"\"Return length of dataset (number of samples)\"\"\"\n",
    "        return self.get_total_number_of_batches()\n",
    "\n",
    "    def calculate_total_number_of_rows(self):\n",
    "        \"\"\"Gets the total number of rows in the data file\"\"\"\n",
    "\n",
    "        row_count = 0\n",
    "\n",
    "        # iterating through the whole file\n",
    "        for row in open(self.filename, \"r\"):\n",
    "            row_count += 1\n",
    "\n",
    "        return row_count\n",
    "\n",
    "    def get_batch_size(self):\n",
    "        \"\"\"Getter for batch size\"\"\"\n",
    "        return self.batch_size\n",
    "\n",
    "    def get_total_number_of_rows(self):\n",
    "        \"\"\"Getter function for total rows\"\"\"\n",
    "        return self.total_rows\n",
    "\n",
    "    def get_rows_to_use(self):\n",
    "        \"\"\"Getter function for rows to use\"\"\"\n",
    "        return self.rows_to_use\n",
    "\n",
    "    def calculate_total_number_of_batches(self):\n",
    "        \"\"\"Number of batches in our dataset\"\"\"\n",
    "\n",
    "        return self.get_total_number_of_rows() // self.get_batch_size()\n",
    "\n",
    "    def get_total_number_of_batches(self):\n",
    "        \"\"\"Getter function for number of total batches\"\"\"\n",
    "        return self.total_num_batches\n",
    "\n",
    "    def get_feature_names(self):\n",
    "        \"\"\"Getter function for feature names\"\"\"\n",
    "        return self.feature_names\n",
    "\n",
    "    def get_prediction_name(self):\n",
    "        \"\"\"Getter for prediction name\"\"\"\n",
    "        return self.prediction_name\n",
    "\n",
    "    def get_current_row(self):\n",
    "        \"\"\"Getter for current row\"\"\"\n",
    "        return self.current_row\n",
    "\n",
    "    def set_reader(self):\n",
    "        \"\"\"Set the csv reader\"\"\"\n",
    "\n",
    "        csvfile = open(self.filename, \"r\")\n",
    "\n",
    "        reader = csv.DictReader(csvfile)\n",
    "\n",
    "        return reader\n",
    "\n",
    "    def get_reader(self):\n",
    "        \"\"\"Get reader\"\"\"\n",
    "        return self.reader\n",
    "\n",
    "    def map_output(self, value: str):\n",
    "        \"\"\"Map to one hot encoding\"\"\"\n",
    "\n",
    "        if value == \"Low\":\n",
    "            return np.array([0.0, 0.0, 1.0])\n",
    "        elif value == \"Medium\":\n",
    "            return np.array([0.0, 1.0, 0.0])\n",
    "        elif value == \"High\":\n",
    "            return np.array([1.0, 0.0, 0.0])\n",
    "        else:\n",
    "            raise Error(f\"Class '{value}' not found\")\n",
    "\n",
    "    def get_next_batch(self):\n",
    "        \"\"\"Get next batch\"\"\"\n",
    "\n",
    "        x_values = []\n",
    "        y_values = []\n",
    "\n",
    "        batch_id = 0\n",
    "\n",
    "        while batch_id < self.get_batch_size():\n",
    "\n",
    "            # Get next row\n",
    "            row = next(self.get_reader())\n",
    "\n",
    "            # Is this row in the rows to use?\n",
    "            if self.get_current_row() in self.get_rows_to_use():\n",
    "                \"\"\"\n",
    "                Note: Here we can pre-process the data however we need to.\n",
    "                For example, we can do random data augmentation\n",
    "                (flips, mirrors, rotations) online as the batch is loaded.\n",
    "                \"\"\"\n",
    "                # Get the feature columns we asked for\n",
    "                values = []\n",
    "\n",
    "                for feature in self.get_feature_names():\n",
    "                    values.append(float(row[feature]))\n",
    "\n",
    "                x_values.append(values)\n",
    "\n",
    "                # Get the prediction columns\n",
    "                y_values.append(self.map_output(row[self.get_prediction_name()]))\n",
    "\n",
    "                batch_id += 1\n",
    "\n",
    "            self.current_row += 1\n",
    "\n",
    "        yield (np.array(x_values), np.array(y_values))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "31bc9d4b",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_loader2 = my_great_data_loader(\n",
    "    filename=csv_filename, batch_size=5, feature_names=feature_names\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6e0bbe91",
   "metadata": {},
   "source": [
    "### __len__()\n",
    "\n",
    "The length command `len()` is actually the function we defined in `__len__()`!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "7159b90a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "121184"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(data_loader2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "857e54be",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "605924"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_loader2.get_total_number_of_rows()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "e6326a13",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "121184"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_loader2.get_total_number_of_batches()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "51f84869",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['pH', 'concentration (μg/ml)', 'molarity (mol/L)',\n",
       "       'temperature (°C)', 'lipophilicity', 'volume (ml)',\n",
       "       'turbidity (NTU)', 'elapsed time (secs)'], dtype='<U21')"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_loader2.get_feature_names()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "113e853f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['pH' 'concentration (μg/ml)' 'molarity (mol/L)' 'temperature (°C)'\n",
      " 'lipophilicity' 'volume (ml)' 'turbidity (NTU)' 'elapsed time (secs)'] values = [[ 5.74404692e+00  7.87489386e-04  5.23464759e+00 -9.38135416e+00\n",
      "   1.25285505e+01  9.36554449e+04  7.63141292e+00  2.34542902e+03]\n",
      " [ 6.19597407e+00  7.31631080e-04  4.92261140e+00  2.07597878e+01\n",
      "   1.16054429e+01  3.18223392e+04  8.66440090e+00  2.36413369e+03]\n",
      " [ 5.40976679e+00  5.03685251e-04  6.28741144e+00  5.72222344e+00\n",
      "   1.54463530e+01  7.78613001e+04  8.61747954e+00  2.36631966e+03]\n",
      " [ 4.78294708e+00  7.70586967e-04  6.51443262e+00 -1.46130240e+00\n",
      "   1.25030559e+01  4.55761131e+04  7.47234089e+00  2.35270893e+03]\n",
      " [ 5.67636363e+00  5.93302441e-04  7.85057187e+00  3.50335152e+00\n",
      "   1.41075041e+01  6.06318252e+04  8.02623528e+00  2.35740530e+03]]\n",
      "X Shape = (5, 8)\n",
      "\n",
      "\n",
      "Antibody affinity values = [[0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]]\n",
      "y Shape = (5, 3)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for x, y in next(data_loader2):\n",
    "\n",
    "    print(f\"{data_loader2.get_feature_names()} values = {x}\")\n",
    "    print(f\"X Shape = {x.shape}\\n\\n\")\n",
    "\n",
    "    print(f\"{data_loader2.get_prediction_name()} values = {y}\")\n",
    "    print(f\"y Shape = {y.shape}\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "fb6b1f1a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['pH' 'concentration (μg/ml)' 'molarity (mol/L)' 'temperature (°C)'\n",
      " 'lipophilicity' 'volume (ml)' 'turbidity (NTU)' 'elapsed time (secs)'] values = [[ 4.54499879e+00  7.88394917e-04  5.48109450e+00 -3.23251874e-01\n",
      "   1.37564326e+01  7.56808245e+04  6.36968383e+00  2.36789287e+03]\n",
      " [ 5.93961608e+00  8.14733582e-04  4.23869546e+00 -2.28936950e+01\n",
      "   1.25375430e+01  8.45087368e+04  8.06484961e+00  2.33552971e+03]\n",
      " [ 6.36593166e+00  2.55069654e-04  5.23936006e+00 -3.12688627e-01\n",
      "   1.74250715e+01  1.16707605e+05  1.09382649e+01  2.35863394e+03]\n",
      " [ 4.29164085e+00  8.35536648e-04  5.65968333e+00 -3.43703584e+01\n",
      "   1.21516149e+01  1.08993641e+05  7.80729313e+00  2.31964426e+03]\n",
      " [ 5.11004783e+00  4.54829158e-04  8.73828694e+00 -1.10449308e+01\n",
      "   1.84803430e+01  5.77676869e+04  6.70575742e+00  2.37577475e+03]]\n",
      "X Shape = (5, 8)\n",
      "\n",
      "\n",
      "Antibody affinity values = [[0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]]\n",
      "y Shape = (5, 3)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for x, y in next(data_loader2):\n",
    "\n",
    "    print(f\"{data_loader2.get_feature_names()} values = {x}\")\n",
    "    print(f\"X Shape = {x.shape}\\n\\n\")\n",
    "\n",
    "    print(f\"{data_loader2.get_prediction_name()} values = {y}\")\n",
    "    print(f\"y Shape = {y.shape}\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "06e94eb1",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['pH' 'concentration (μg/ml)' 'molarity (mol/L)' 'temperature (°C)'\n",
      " 'lipophilicity' 'volume (ml)' 'turbidity (NTU)' 'elapsed time (secs)'] values = [[ 6.25537570e+00  6.43369972e-04  6.56668552e+00  2.32853701e+01\n",
      "   1.37433971e+01  7.01466670e+04  7.35643303e+00  2.37831048e+03]\n",
      " [ 6.48863554e+00  8.94117682e-04  5.31515895e+00 -4.93134078e+00\n",
      "   1.20145251e+01  8.57713255e+04  6.38796320e+00  2.35397101e+03]\n",
      " [ 4.16911609e+00  1.06951387e-03  5.51811419e+00  4.30834299e+00\n",
      "   8.88748948e+00  9.70224800e+04  5.93197832e+00  2.34516033e+03]\n",
      " [ 5.50827091e+00  4.02971249e-04  5.71758970e+00  9.23725690e-01\n",
      "   1.61043008e+01  9.11110962e+04  9.81191369e+00  2.35958439e+03]\n",
      " [ 5.75426364e+00  5.84485355e-04  5.72744424e+00 -6.78131027e+00\n",
      "   1.43045115e+01  7.56740047e+04  8.97553103e+00  2.34943438e+03]]\n",
      "X Shape = (5, 8)\n",
      "\n",
      "\n",
      "Antibody affinity values = [[0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]]\n",
      "y Shape = (5, 3)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for x, y in next(data_loader2):\n",
    "\n",
    "    print(f\"{data_loader2.get_feature_names()} values = {x}\")\n",
    "    print(f\"X Shape = {x.shape}\\n\\n\")\n",
    "\n",
    "    print(f\"{data_loader2.get_prediction_name()} values = {y}\")\n",
    "    print(f\"y Shape = {y.shape}\\n\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "ac250213",
   "metadata": {},
   "source": [
    "### We can even get a data loader with just some of our features"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "d19d4228",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_loader3 = my_great_data_loader(\n",
    "    filename=csv_filename, batch_size=5, feature_names=[\"pH\", \"turbidity (NTU)\"]\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "9b52e704",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['pH', 'turbidity (NTU)']"
      ]
     },
     "execution_count": 42,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_loader3.get_feature_names()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "1f35d7c6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['pH', 'turbidity (NTU)'] values = [[5.74404692 7.63141292]\n",
      " [6.19597407 8.6644009 ]\n",
      " [5.40976679 8.61747954]\n",
      " [4.78294708 7.47234089]\n",
      " [5.67636363 8.02623528]]\n",
      "X Shape = (5, 2)\n",
      "\n",
      "\n",
      "Antibody affinity values = [[0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]]\n",
      "y Shape = (5, 3)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for x, y in next(data_loader3):\n",
    "\n",
    "    print(f\"{data_loader3.get_feature_names()} values = {x}\")\n",
    "    print(f\"X Shape = {x.shape}\\n\\n\")\n",
    "\n",
    "    print(f\"{data_loader3.get_prediction_name()} values = {y}\")\n",
    "    print(f\"y Shape = {y.shape}\\n\\n\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9bd5e177",
   "metadata": {},
   "source": [
    "### And with just specific rows of the CSV file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "694f715f",
   "metadata": {},
   "outputs": [],
   "source": [
    "data_loader4 = my_great_data_loader(\n",
    "    filename=csv_filename,\n",
    "    batch_size=5,\n",
    "    feature_names=[\"pH\", \"turbidity (NTU)\"],\n",
    "    rows_to_use=[14, 223, 1234, 2222, 20222, 45123, 83456],\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "fd24fdd5",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['pH', 'turbidity (NTU)'] values = [[5.75426364 8.97553103]\n",
      " [6.65077974 2.21735753]\n",
      " [5.9008641  9.58502736]\n",
      " [3.39149966 7.6406676 ]\n",
      " [7.19851772 7.65204959]]\n",
      "X Shape = (5, 2)\n",
      "\n",
      "\n",
      "Antibody affinity values = [[0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]]\n",
      "y Shape = (5, 3)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for x, y in next(data_loader4):\n",
    "\n",
    "    print(f\"{data_loader4.get_feature_names()} values = {x}\")\n",
    "    print(f\"X Shape = {x.shape}\\n\\n\")\n",
    "\n",
    "    print(f\"{data_loader4.get_prediction_name()} values = {y}\")\n",
    "    print(f\"y Shape = {y.shape}\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "bbe218ee",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[14, 223, 1234, 2222, 20222, 45123, 83456]"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data_loader4.get_rows_to_use()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "bbfe4ddd",
   "metadata": {},
   "source": [
    "### Let's do our own train/test/split of the data\n",
    "\n",
    "To complete our data loader, we need to be able to [split the dataset](https://scikit-learn.org/stable/modules/generated/sklearn.model_selection.train_test_split.html) into 3 parts: a training set, a testing set, and a validation set. Usually the split is 80/10/10 (0.8/0.1/0.1). We'll do this by:\n",
    "\n",
    "* creating lists of the indices\n",
    "* randomizing that list\n",
    "* split the list into 3 sections at the desired indices\n",
    "* return the 3 lists (individually sorted)\n",
    "\n",
    "NOTE: These lists are just the indices (row numbers) not the actual data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "489b365d",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_splits(\n",
    "    filename,\n",
    "    train_test_split: float = 0.8,\n",
    "    test_validation_split: float = 0.5,\n",
    "    random_seed: int = 816,\n",
    "):\n",
    "    \"\"\"Calculates the indices for the train/test/validation split\n",
    "\n",
    "    Args:\n",
    "        filename(str): The CSV filename\n",
    "        train_test_split(float): Number from 0 to 1 (Default 0.8 which is 80% split)\n",
    "        test_validation_split(float): Number from 0 to 1 (Default 0.5 which is 50% split)\n",
    "        random_seed(int): Random seed\n",
    "\n",
    "    \"\"\"\n",
    "\n",
    "    # Make sure these are percentages\n",
    "    assert train_test_split >= 0\n",
    "    assert train_test_split <= 1\n",
    "    assert test_validation_split >= 0\n",
    "    assert test_validation_split <= 1\n",
    "\n",
    "    total_rows = 0\n",
    "\n",
    "    # iterating through the whole file\n",
    "    for row in open(filename, \"r\"):\n",
    "        total_rows += 1\n",
    "\n",
    "    total_indicies = np.arange(total_rows)\n",
    "\n",
    "    np.random.seed(\n",
    "        random_seed\n",
    "    )  # Make sure we get the same random sequence whenever we run it\n",
    "    np.random.shuffle(\n",
    "        total_indicies\n",
    "    )  # Recall we seeded so this should be deterministic\n",
    "\n",
    "    last_train_idx = int(train_test_split * total_rows)\n",
    "\n",
    "    train_rows = np.sort(total_indicies[:last_train_idx])\n",
    "    remaining_rows = total_indicies[last_train_idx:]\n",
    "\n",
    "    last_validation_idx = int(test_validation_split * len(remaining_rows))\n",
    "    validation_rows = np.sort(remaining_rows[:last_validation_idx])\n",
    "    test_rows = np.sort(remaining_rows[last_validation_idx:])\n",
    "\n",
    "    return train_rows, test_rows, validation_rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "a548b94e",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_rows, test_rows, validation_rows = get_splits(csv_filename)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "5421a84f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "484739"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "1c27c7b5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "60593"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(test_rows)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "5a06cf2a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "60592"
      ]
     },
     "execution_count": 51,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(validation_rows)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "dbe9ca15",
   "metadata": {},
   "source": [
    "### If we did it correctly, then each list is unique"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "1324fcbf",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([], dtype=int64)"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.intersect1d(train_rows, test_rows)  # There should be no overlapping rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "990ee65c",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([], dtype=int64)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.intersect1d(validation_rows, test_rows)  # There should be no overlapping rows"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "f53db650",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([], dtype=int64)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.intersect1d(train_rows, validation_rows)  # There should be no overlapping rows"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "9b180e09",
   "metadata": {},
   "source": [
    "### Finally we can create 3 separate data loaders\n",
    "\n",
    "One for training data, one for testing data, and one for validation data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "7e8c1596",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_loader = my_great_data_loader(\n",
    "    filename=csv_filename,\n",
    "    batch_size=32,\n",
    "    rows_to_use=train_rows,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "39f7ba61",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_loader = my_great_data_loader(\n",
    "    filename=csv_filename,\n",
    "    batch_size=16,\n",
    "    rows_to_use=test_rows,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "b09800b2",
   "metadata": {},
   "outputs": [],
   "source": [
    "validation_loader = my_great_data_loader(\n",
    "    filename=csv_filename,\n",
    "    batch_size=8,\n",
    "    rows_to_use=validation_rows,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "e01452ea",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "['pH' 'concentration (μg/ml)' 'molarity (mol/L)' 'temperature (°C)'\n",
      " 'lipophilicity' 'volume (ml)' 'turbidity (NTU)' 'elapsed time (secs)'] values = [[ 5.67636363e+00  5.93302441e-04  7.85057187e+00  3.50335152e+00\n",
      "   1.41075041e+01  6.06318252e+04  8.02623528e+00  2.35740530e+03]\n",
      " [ 4.54499879e+00  7.88394917e-04  5.48109450e+00 -3.23251874e-01\n",
      "   1.37564326e+01  7.56808245e+04  6.36968383e+00  2.36789287e+03]\n",
      " [ 5.93961608e+00  8.14733582e-04  4.23869546e+00 -2.28936950e+01\n",
      "   1.25375430e+01  8.45087368e+04  8.06484961e+00  2.33552971e+03]\n",
      " [ 6.36593166e+00  2.55069654e-04  5.23936006e+00 -3.12688627e-01\n",
      "   1.74250715e+01  1.16707605e+05  1.09382649e+01  2.35863394e+03]\n",
      " [ 4.29164085e+00  8.35536648e-04  5.65968333e+00 -3.43703584e+01\n",
      "   1.21516149e+01  1.08993641e+05  7.80729313e+00  2.31964426e+03]\n",
      " [ 5.11004783e+00  4.54829158e-04  8.73828694e+00 -1.10449308e+01\n",
      "   1.84803430e+01  5.77676869e+04  6.70575742e+00  2.37577475e+03]\n",
      " [ 6.25537570e+00  6.43369972e-04  6.56668552e+00  2.32853701e+01\n",
      "   1.37433971e+01  7.01466670e+04  7.35643303e+00  2.37831048e+03]\n",
      " [ 6.48863554e+00  8.94117682e-04  5.31515895e+00 -4.93134078e+00\n",
      "   1.20145251e+01  8.57713255e+04  6.38796320e+00  2.35397101e+03]\n",
      " [ 4.16911609e+00  1.06951387e-03  5.51811419e+00  4.30834299e+00\n",
      "   8.88748948e+00  9.70224800e+04  5.93197832e+00  2.34516033e+03]\n",
      " [ 5.50827091e+00  4.02971249e-04  5.71758970e+00  9.23725690e-01\n",
      "   1.61043008e+01  9.11110962e+04  9.81191369e+00  2.35958439e+03]\n",
      " [ 5.75426364e+00  5.84485355e-04  5.72744424e+00 -6.78131027e+00\n",
      "   1.43045115e+01  7.56740047e+04  8.97553103e+00  2.34943438e+03]\n",
      " [ 4.80553931e+00  5.32877457e-04  4.00164694e+00  2.23096795e+01\n",
      "   1.22745684e+01  9.39674873e+04  1.08917854e+01  2.35361041e+03]\n",
      " [ 5.10101329e+00  7.67520387e-04  8.09306676e+00  2.69005516e+01\n",
      "   1.10617599e+01  6.79461979e+04  7.07817771e+00  2.36363652e+03]\n",
      " [ 5.84355098e+00  6.26901400e-04  6.65636799e+00 -1.50878689e+00\n",
      "   1.19334438e+01  8.24597168e+04  9.72358799e+00  2.33271230e+03]\n",
      " [ 2.92440776e+00  8.17825429e-04  4.11655812e+00 -1.58603353e+01\n",
      "   1.28707369e+01  9.14419323e+04  7.53297715e+00  2.34608294e+03]\n",
      " [ 5.55006916e+00  4.98359936e-04  5.29002294e+00 -1.20161493e+01\n",
      "   1.54278402e+01  7.30114247e+04  9.65035159e+00  2.34892897e+03]\n",
      " [ 5.08518290e+00  7.07904030e-04  8.53336018e+00 -1.70380312e+01\n",
      "   1.50760255e+01  6.18195905e+04  6.02831095e+00  2.35614997e+03]\n",
      " [ 3.75511822e+00  6.20212111e-04  7.99495623e+00  1.25660100e+01\n",
      "   1.27751629e+01  7.56639475e+04  8.27385463e+00  2.35418579e+03]\n",
      " [ 6.79350995e+00  7.37581408e-04  7.96280461e+00 -1.66846072e+01\n",
      "   1.32742815e+01  5.28938633e+04  7.27423309e+00  2.34119119e+03]\n",
      " [ 5.72513839e+00  7.46088614e-04  7.85749435e+00  6.56326795e+00\n",
      "   1.42306196e+01  7.92405470e+04  5.53275606e+00  2.37403715e+03]\n",
      " [ 5.27291525e+00  4.03175544e-04  8.01585596e+00  1.07707890e+01\n",
      "   1.42618102e+01  7.56826325e+04  1.02500955e+01  2.34918171e+03]\n",
      " [ 6.76366550e+00  6.85966007e-04  7.50098686e+00  1.37499105e+01\n",
      "   1.37563622e+01  5.17979450e+04  6.82243821e+00  2.37261866e+03]\n",
      " [ 4.89393702e+00  8.28516961e-04  5.08374885e+00 -1.79290703e+01\n",
      "   1.20916042e+01  9.57788585e+04  7.70510243e+00  2.33589243e+03]\n",
      " [ 4.18389433e+00  6.96726664e-04  7.70290659e+00  1.40531520e+01\n",
      "   1.38474880e+01  7.02950691e+04  6.40975622e+00  2.37382846e+03]\n",
      " [ 4.41929970e+00  8.65897277e-04  6.96134891e+00 -1.93285251e+00\n",
      "   1.36847986e+01  6.31131166e+04  4.92659719e+00  2.37156196e+03]\n",
      " [ 5.49448713e+00  5.99417838e-04  8.51201180e+00  2.45420525e+01\n",
      "   1.46213943e+01  6.61800492e+04  6.59500970e+00  2.38354075e+03]\n",
      " [ 4.39768003e+00  8.38860204e-04  6.76539935e+00  1.12166572e+01\n",
      "   1.33908377e+01  8.22714662e+04  5.08461374e+00  2.37866786e+03]\n",
      " [ 5.07963596e+00  6.28395131e-04  7.24814244e+00  2.11443581e+00\n",
      "   1.50407432e+01  8.25208066e+04  6.90916903e+00  2.36823795e+03]\n",
      " [ 4.54571278e+00  5.23071176e-04  8.03177333e+00  1.41672209e+01\n",
      "   1.37078850e+01  8.39573828e+04  8.81164396e+00  2.35674415e+03]\n",
      " [ 5.32302295e+00  5.90643712e-04  8.31536063e+00 -1.46069276e+01\n",
      "   1.47395311e+01  7.62418029e+04  7.89034867e+00  2.34457749e+03]\n",
      " [ 4.95535871e+00  3.80316158e-04  8.42556416e+00  1.32611059e+01\n",
      "   1.43202487e+01  9.63518138e+04  1.01903174e+01  2.34909706e+03]\n",
      " [ 6.48091304e+00  8.30889623e-04  4.80774228e+00 -4.17255351e+00\n",
      "   1.29796480e+01  9.20347029e+04  6.64180011e+00  2.35945092e+03]]\n",
      "X Shape = (32, 8)\n",
      "\n",
      "\n",
      "Antibody affinity values = [[0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]\n",
      " [0. 0. 1.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 0. 1.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [1. 0. 0.]\n",
      " [0. 1. 0.]\n",
      " [0. 0. 1.]]\n",
      "y Shape = (32, 3)\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for x, y in next(train_loader):\n",
    "\n",
    "    print(f\"{train_loader.get_feature_names()} values = {x}\")\n",
    "    print(f\"X Shape = {x.shape}\\n\\n\")\n",
    "\n",
    "    print(f\"{train_loader.get_prediction_name()} values = {y}\")\n",
    "    print(f\"y Shape = {y.shape}\\n\\n\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "f397514e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "18935"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(train_loader)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "df0b6a1d",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.10.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}