performance_comparison.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Comparison of models' performances\n",
    "\n",
    "I will load three trained models I made available for each kind, and will inspect their errors statistics produced on the Test set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import time\n",
    "import yaml\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn\n",
    "\n",
    "# local imports\n",
    "import tools, holdout"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load config params\n",
    "params = yaml.load( open(os.getcwd() + '/config.yaml'), yaml.Loader ) \n",
    "\n",
    "# Add this for simplicity\n",
    "params['size_nan'] = int(params['len_input']*params['total_nan_share']) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Setting GPU configurations.\n",
      "1 Physical GPUs, 1 Logical GPUs\n"
     ]
    }
   ],
   "source": [
    "# Once params are loaded, load TensorFlow and set its GPU config\n",
    "import tensorflow as tf\n",
    "tools.set_gpu_configurations(params) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:No training configuration found in save file: the model was *not* compiled. Compile it manually.\n",
      "WARNING:tensorflow:No training configuration found in save file: the model was *not* compiled. Compile it manually.\n",
      "WARNING:tensorflow:No training configuration found in save file: the model was *not* compiled. Compile it manually.\n"
     ]
    }
   ],
   "source": [
    "# Load models by name\n",
    "seq2seq_name = 'seq2seq_00'\n",
    "GAN_name = 'GAN_00'\n",
    "pGAN_name = 'pGAN_00'\n",
    "\n",
    "seq2seq = tf.keras.models.load_model(os.getcwd() + '/saved_models/' + seq2seq_name +'.h5') \n",
    "GAN = tf.keras.models.load_model(os.getcwd() + '/saved_models/' + GAN_name +'.h5') \n",
    "pGAN = tf.keras.models.load_model(os.getcwd() + '/saved_models/' + pGAN_name +'.h5') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Run models on Test set\n",
    "\n",
    "Normally, `holdout.run_test()` returns the following arrays: `X` and `Y` data, predictions, errors, together with a dictionary of error statistics. In this Notebook I will focus only on error arrays `E_` and error `stats_`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Check model performance on Validation data.\n",
      "Done in 307.11ss.\n",
      "\n",
      "\tMAE Loss: 0.05882476270198822\n",
      "\n",
      "Error statistics:\n",
      "\tMean:            0.05882476\n",
      "\tSt dev:          0.02463975 \n",
      "\n",
      "\tMin:             0.0060952753\n",
      "\t25th percentile: 0.04234670661389828\n",
      "\tMedian:          0.055126525\n",
      "\t75th percentile: 0.07095329836010933\n",
      "\tMax:             0.38657707\n",
      "\n",
      "\n",
      "Check model performance on Test data.\n",
      "Done in 299.02ss.\n",
      "\n",
      "\tMAE Loss: 0.05898692086338997\n",
      "\n",
      "Error statistics:\n",
      "\tMean:            0.058986932\n",
      "\tSt dev:          0.0248263 \n",
      "\n",
      "\tMin:             0.0029453349\n",
      "\t25th percentile: 0.04276513680815697\n",
      "\tMedian:          0.05495383\n",
      "\t75th percentile: 0.07066602632403374\n",
      "\tMax:             0.43441105\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Seq2seq\n",
    "\n",
    "_, _, _, E_seq2seq, stats_seq2seq = holdout.run_test(\n",
    "    model = seq2seq, \n",
    "    params = params, \n",
    "    check_test_performance = True, \n",
    "    return_stats = True) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Check model performance on Validation data.\n",
      "Done in 302.59ss.\n",
      "\n",
      "\tMAE Loss: 0.5480108857154846\n",
      "\n",
      "Error statistics:\n",
      "\tMean:            0.5480094\n",
      "\tSt dev:          0.21135655 \n",
      "\n",
      "\tMin:             0.0\n",
      "\t25th percentile: 0.3788064271211624\n",
      "\tMedian:          0.57117313\n",
      "\t75th percentile: 0.7102163434028625\n",
      "\tMax:             1.2848921\n",
      "\n",
      "\n",
      "Check model performance on Test data.\n",
      "Done in 303.67ss.\n",
      "\n",
      "\tMAE Loss: 0.5487329959869385\n",
      "\n",
      "Error statistics:\n",
      "\tMean:            0.54873157\n",
      "\tSt dev:          0.21106702 \n",
      "\n",
      "\tMin:             0.0\n",
      "\t25th percentile: 0.38072822988033295\n",
      "\tMedian:          0.57197475\n",
      "\t75th percentile: 0.7085483074188232\n",
      "\tMax:             1.2915955\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# GAN's Generator\n",
    "\n",
    "_, _, _, E_gan, stats_gan = holdout.run_test(\n",
    "    model = GAN, \n",
    "    params = params, \n",
    "    check_test_performance = True, \n",
    "    return_stats = True) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Check model performance on Validation data.\n",
      "Done in 322.21ss.\n",
      "\n",
      "\tMAE Loss: 0.04670098423957825\n",
      "\n",
      "Error statistics:\n",
      "\tMean:            0.046700977\n",
      "\tSt dev:          0.0157721 \n",
      "\n",
      "\tMin:             0.002786517\n",
      "\t25th percentile: 0.035480765625834465\n",
      "\tMedian:          0.045017138\n",
      "\t75th percentile: 0.05553138628602028\n",
      "\tMax:             0.20436941\n",
      "\n",
      "\n",
      "Check model performance on Test data.\n",
      "Done in 313.27ss.\n",
      "\n",
      "\tMAE Loss: 0.0466911755502224\n",
      "\n",
      "Error statistics:\n",
      "\tMean:            0.046691168\n",
      "\tSt dev:          0.015540862 \n",
      "\n",
      "\tMin:             0.0029406825\n",
      "\t25th percentile: 0.03535008616745472\n",
      "\tMedian:          0.04525705\n",
      "\t75th percentile: 0.05576721020042896\n",
      "\tMax:             0.18967311\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Partial GAN's Generator\n",
    "\n",
    "_, _, _, E_pgan, stats_pgan = holdout.run_test(\n",
    "    model = pGAN, \n",
    "    params = params, \n",
    "    check_test_performance = True, \n",
    "    return_stats = True) "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's take a look at how error distributions compare to each other:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,5))\n",
    "plt.title('Test error distributions of the three models implemented (Mean Absolute Error).')\n",
    "seaborn.kdeplot(E_seq2seq, shade=True, label='Vanilla Seq2seq')\n",
    "seaborn.kdeplot(E_gan, shade=True, label='GAN')\n",
    "seaborn.kdeplot(E_pgan, shade=True, label='partial GAN')\n",
    "plt.show() "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "It seems clear that partial GAN is the best model. The Discriminator's feedback provided a slight improvement over the vanilla Seq2seq model, which is its real competitor. A plain, canonical GAN seems insufficient.\n",
    "\n",
    "Let's take a closer look at the two best models:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,5))\n",
    "plt.title('Vanilla Se2seq VS partial GAN on Test error (Mean Absolute Error).')\n",
    "seaborn.kdeplot(E_seq2seq, shade=True, label='Vanilla Seq2seq')\n",
    "seaborn.kdeplot(E_pgan, shade=True, label='partial GAN')\n",
    "plt.show() "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}