RMSE_figures.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.ticker as ticker\n",
    "\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, MaxPooling2D, Activation, Dropout\n",
    "from keras.layers import Flatten, Input, Dense, BatchNormalization\n",
    "from keras import backend as K\n",
    "from keras.utils import multi_gpu_model, to_categorical, plot_model\n",
    "\n",
    "import keras\n",
    "import tensorflow as tf\n",
    "from sklearn.model_selection import train_test_split\n",
    "\n",
    "import os, sys, errno\n",
    "os.environ['HDF5_USE_FILE_LOCKING'] = 'FALSE'\n",
    "os.environ['CUDA_VISIBLE_DEVICES'] = '0'\n",
    "\n",
    "import csv\n",
    "import codecs\n",
    "import pickle\n",
    "import numpy as np\n",
    "import glob\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "\n",
    "import train_cnn\n",
    "\n",
    "# Global values\n",
    "PATH = '/datax/scratch/bbrzycki/data/nb-localization'\n",
    "TCHANS, FCHANS = 32, 1024\n",
    "\n",
    "experiment_path = '/datax/scratch/bbrzycki/data/nb-localization/training/'\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## One signal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "d = {'db': [], 'rmse_index': [], 'model': []}\n",
    "\n",
    "pred_dicts = {\n",
    "    'Baseline': np.load(experiment_path + 'baseline_1sig_32bs/test_predictions.npy',\n",
    "                        allow_pickle=True).item(),\n",
    "    'Baseline Bright': np.load(experiment_path + 'baseline_1sig_32bs_bright/test_predictions.npy',\n",
    "                               allow_pickle=True).item(),\n",
    "    'Final': np.load(experiment_path + 'final_1sig_32bs/test_predictions.npy',\n",
    "                     allow_pickle=True).item(),\n",
    "    'Final Bright': np.load(experiment_path + 'final_1sig_32bs_bright/test_predictions.npy',\n",
    "                            allow_pickle=True).item(),\n",
    "}\n",
    "\n",
    "for db in range(0, 30, 5):\n",
    "    for j in range(4000):\n",
    "        fn = '{:02}db_{:06d}.npy'.format(db, j)\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Baseline'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Baseline')\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Baseline Bright'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Baseline Bright')\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Final'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Final')\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Final Bright'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Final Bright')\n",
    "\n",
    "df = pd.DataFrame(data=d)\n",
    "\n",
    "sns.set(palette=\"rainbow\")\n",
    "fig = plt.figure(figsize=(10, 8))\n",
    "ax = sns.boxplot(x='db', y='rmse_index', hue='model', data=df, saturation=1, fliersize=2)\n",
    "h, l = ax.get_legend_handles_labels()\n",
    "ax.legend(h, l, loc=3)\n",
    "\n",
    "plt.yscale('log')\n",
    "plt.xlabel('Signal Intensity (dB)')\n",
    "plt.ylabel('RMSE (index units)')\n",
    "\n",
    "ax.set_yticks([1, 3, 10, 30, 100, 300, 1000])\n",
    "ax.yaxis.set_major_formatter(ticker.ScalarFormatter())\n",
    "\n",
    "# plt.savefig('1sig_rmse.pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Two signals"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "d = {'db': [], 'rmse_index': [], 'model': []}\n",
    "\n",
    "pred_dicts = {\n",
    "    'Baseline': np.load(experiment_path + 'baseline_2sig_32bs/test_predictions.npy',\n",
    "                        allow_pickle=True).item(),\n",
    "    'Baseline Bright': np.load(experiment_path + 'baseline_2sig_32bs_bright/test_predictions.npy',\n",
    "                               allow_pickle=True).item(),\n",
    "    'Final': np.load(experiment_path + 'final_2sig_32bs/test_predictions.npy',\n",
    "                     allow_pickle=True).item(),\n",
    "    'Final Bright': np.load(experiment_path + 'final_2sig_32bs_bright/test_predictions.npy',\n",
    "                            allow_pickle=True).item(),\n",
    "}\n",
    "\n",
    "for db in range(0, 30, 5):\n",
    "    for j in range(4000):\n",
    "        fn = '{:02}db_{:06d}.npy'.format(db, j)\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Baseline'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Baseline')\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Baseline Bright'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Baseline Bright')\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Final'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Final')\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Final Bright'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Final Bright')\n",
    "\n",
    "df = pd.DataFrame(data=d)\n",
    "\n",
    "sns.set(palette=\"rainbow\")\n",
    "fig = plt.figure(figsize=(10, 8))\n",
    "ax = sns.boxplot(x='db', y='rmse_index', hue='model', data=df, saturation=1, fliersize=2)\n",
    "h, l = ax.get_legend_handles_labels()\n",
    "ax.legend(h, l, loc=3)\n",
    "\n",
    "plt.yscale('log')\n",
    "plt.xlabel('Signal Intensity (dB)')\n",
    "plt.ylabel('RMSE (index units)')\n",
    "\n",
    "ax.set_yticks([1, 3, 10, 30, 100, 300, 1000])\n",
    "ax.yaxis.set_major_formatter(ticker.ScalarFormatter())\n",
    "\n",
    "# plt.savefig('2sig_rmse.pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Compare Final to TurboSETI"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "d = {'db': [], 'rmse_index': [], 'model': []}\n",
    "\n",
    "pred_dicts = {\n",
    "    'Final Bright': np.load(experiment_path + 'final_1sig_32bs_bright/test_predictions.npy',\n",
    "                            allow_pickle=True).item(),\n",
    "    'TurboSETI Bright': np.load(experiment_path + 'turboseti/test_predictions.npy',\n",
    "                            allow_pickle=True).item(),\n",
    "}\n",
    "\n",
    "for db in range(10, 30, 5):\n",
    "    for j in range(4000):\n",
    "        fn = '{:02}db_{:06d}.npy'.format(db, j)\n",
    "        \n",
    "        d['db'].append(db)\n",
    "        lb = pred_dicts['Final Bright'][fn]\n",
    "        rmse_index = np.mean((lb[0] - lb[1])**2)**0.5 * 1024\n",
    "        d['rmse_index'].append(rmse_index)\n",
    "        d['model'].append('Final Bright')\n",
    "        \n",
    "        if fn in pred_dicts['TurboSETI Bright']:\n",
    "            # For some cases, turboseti doesn't find any signals\n",
    "            d['db'].append(db)\n",
    "            rmse_index = pred_dicts['TurboSETI Bright'][fn]\n",
    "            d['rmse_index'].append(rmse_index)\n",
    "            d['model'].append('TurboSETI Bright')\n",
    "        \n",
    "\n",
    "df = pd.DataFrame(data=d)\n",
    "\n",
    "sns.set(palette=\"rainbow\")\n",
    "current_palette = sns.color_palette()\n",
    "\n",
    "fig = plt.figure(figsize=(10, 8))\n",
    "my_pal = {\n",
    "    'Final Bright': current_palette[3],\n",
    "    'TurboSETI Bright': current_palette[4]\n",
    "}\n",
    "ax = sns.boxplot(x='db', \n",
    "                 y='rmse_index',\n",
    "                 hue='model', \n",
    "                 data=df,\n",
    "                 saturation=1, \n",
    "                 fliersize=2,\n",
    "                 palette=my_pal)\n",
    "h, l = ax.get_legend_handles_labels()\n",
    "ax.legend(h, l, loc=1)\n",
    "\n",
    "plt.yscale('log')\n",
    "plt.xlabel('Signal Intensity (dB)')\n",
    "plt.ylabel('RMSE (index units)')\n",
    "\n",
    "ax.set_yticks([1, 3, 10, 30, 100, 300, 1000])\n",
    "ax.yaxis.set_major_formatter(ticker.ScalarFormatter())\n",
    "plt.ylim(0.03, 1000)\n",
    "\n",
    "# plt.savefig('turboseti_rmse.pdf', bbox_inches='tight')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python (bl_tf)",
   "language": "python",
   "name": "ml"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}