03_Dropout.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<center><div style=\"direction:rtl;font-family:B Lotus, B Nazanin, Tahoma\">به نام خدا</div></center>\n",
    "\n",
    "<h1><center><div style=\"direction:rtl;font-family:B Lotus, B Nazanin, Tahoma\">مشکل بیش‌برازش داده‌ها یا Overfitting و معرفی Dropout</div></center></h1>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">شبکه بدون Dropout</div>\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/150\n",
      "8/8 [==============================] - 22s 104ms/step - loss: 2.2233 - accuracy: 0.1861 - val_loss: 2.0797 - val_accuracy: 0.3400\n",
      "Epoch 2/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.9889 - accuracy: 0.4697 - val_loss: 1.8246 - val_accuracy: 0.6200\n",
      "Epoch 3/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.6978 - accuracy: 0.6735 - val_loss: 1.5084 - val_accuracy: 0.6850\n",
      "Epoch 4/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.3603 - accuracy: 0.7463 - val_loss: 1.1714 - val_accuracy: 0.7800\n",
      "Epoch 5/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.0428 - accuracy: 0.8013 - val_loss: 0.8813 - val_accuracy: 0.8300\n",
      "Epoch 6/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.7607 - accuracy: 0.8501 - val_loss: 0.6759 - val_accuracy: 0.8550\n",
      "Epoch 7/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.6057 - accuracy: 0.8611 - val_loss: 0.5510 - val_accuracy: 0.8700\n",
      "Epoch 8/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.5054 - accuracy: 0.8697 - val_loss: 0.4830 - val_accuracy: 0.8900\n",
      "Epoch 9/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.4135 - accuracy: 0.8854 - val_loss: 0.4292 - val_accuracy: 0.8950\n",
      "Epoch 10/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.3829 - accuracy: 0.8899 - val_loss: 0.3918 - val_accuracy: 0.8850\n",
      "Epoch 11/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3477 - accuracy: 0.8962 - val_loss: 0.3705 - val_accuracy: 0.8900\n",
      "Epoch 12/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.2944 - accuracy: 0.9196 - val_loss: 0.3507 - val_accuracy: 0.8950\n",
      "Epoch 13/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3071 - accuracy: 0.9093 - val_loss: 0.3287 - val_accuracy: 0.9000\n",
      "Epoch 14/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2809 - accuracy: 0.9166 - val_loss: 0.3183 - val_accuracy: 0.9000\n",
      "Epoch 15/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2773 - accuracy: 0.9023 - val_loss: 0.3108 - val_accuracy: 0.8900\n",
      "Epoch 16/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2543 - accuracy: 0.9236 - val_loss: 0.3010 - val_accuracy: 0.8850\n",
      "Epoch 17/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2520 - accuracy: 0.9163 - val_loss: 0.2945 - val_accuracy: 0.9000\n",
      "Epoch 18/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2355 - accuracy: 0.9339 - val_loss: 0.2872 - val_accuracy: 0.9000\n",
      "Epoch 19/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2300 - accuracy: 0.9362 - val_loss: 0.2826 - val_accuracy: 0.8950\n",
      "Epoch 20/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2155 - accuracy: 0.9200 - val_loss: 0.2767 - val_accuracy: 0.8950\n",
      "Epoch 21/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2063 - accuracy: 0.9399 - val_loss: 0.2704 - val_accuracy: 0.9050\n",
      "Epoch 22/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2027 - accuracy: 0.9384 - val_loss: 0.2703 - val_accuracy: 0.9050\n",
      "Epoch 23/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1910 - accuracy: 0.9430 - val_loss: 0.2690 - val_accuracy: 0.9000\n",
      "Epoch 24/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1892 - accuracy: 0.9430 - val_loss: 0.2598 - val_accuracy: 0.9150\n",
      "Epoch 25/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1743 - accuracy: 0.9485 - val_loss: 0.2639 - val_accuracy: 0.9000\n",
      "Epoch 26/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.1633 - accuracy: 0.9469 - val_loss: 0.2592 - val_accuracy: 0.9100\n",
      "Epoch 27/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1595 - accuracy: 0.9532 - val_loss: 0.2522 - val_accuracy: 0.9150\n",
      "Epoch 28/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1629 - accuracy: 0.9457 - val_loss: 0.2552 - val_accuracy: 0.9100\n",
      "Epoch 29/150\n",
      "8/8 [==============================] - 0s 11ms/step - loss: 0.1431 - accuracy: 0.9539 - val_loss: 0.2521 - val_accuracy: 0.9100\n",
      "Epoch 30/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.1462 - accuracy: 0.9519 - val_loss: 0.2479 - val_accuracy: 0.9100\n",
      "Epoch 31/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1391 - accuracy: 0.9514 - val_loss: 0.2470 - val_accuracy: 0.9050\n",
      "Epoch 32/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.1346 - accuracy: 0.9551 - val_loss: 0.2489 - val_accuracy: 0.9150\n",
      "Epoch 33/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1384 - accuracy: 0.9541 - val_loss: 0.2488 - val_accuracy: 0.9200\n",
      "Epoch 34/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1409 - accuracy: 0.9491 - val_loss: 0.2446 - val_accuracy: 0.9200\n",
      "Epoch 35/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.1453 - accuracy: 0.9549 - val_loss: 0.2451 - val_accuracy: 0.9250\n",
      "Epoch 36/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1172 - accuracy: 0.9666 - val_loss: 0.2429 - val_accuracy: 0.9200\n",
      "Epoch 37/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1141 - accuracy: 0.9649 - val_loss: 0.2425 - val_accuracy: 0.9250\n",
      "Epoch 38/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1265 - accuracy: 0.9592 - val_loss: 0.2466 - val_accuracy: 0.9300\n",
      "Epoch 39/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1248 - accuracy: 0.9636 - val_loss: 0.2382 - val_accuracy: 0.9350\n",
      "Epoch 40/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1183 - accuracy: 0.9583 - val_loss: 0.2449 - val_accuracy: 0.9250\n",
      "Epoch 41/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1072 - accuracy: 0.9679 - val_loss: 0.2398 - val_accuracy: 0.9300\n",
      "Epoch 42/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1208 - accuracy: 0.9657 - val_loss: 0.2419 - val_accuracy: 0.9300\n",
      "Epoch 43/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1026 - accuracy: 0.9735 - val_loss: 0.2417 - val_accuracy: 0.9250\n",
      "Epoch 44/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0999 - accuracy: 0.9784 - val_loss: 0.2444 - val_accuracy: 0.9350\n",
      "Epoch 45/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1073 - accuracy: 0.9718 - val_loss: 0.2446 - val_accuracy: 0.9200\n",
      "Epoch 46/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1033 - accuracy: 0.9739 - val_loss: 0.2344 - val_accuracy: 0.9350\n",
      "Epoch 47/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0935 - accuracy: 0.9724 - val_loss: 0.2411 - val_accuracy: 0.9300\n",
      "Epoch 48/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0823 - accuracy: 0.9808 - val_loss: 0.2404 - val_accuracy: 0.9350\n",
      "Epoch 49/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0844 - accuracy: 0.9787 - val_loss: 0.2415 - val_accuracy: 0.9300\n",
      "Epoch 50/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0755 - accuracy: 0.9847 - val_loss: 0.2420 - val_accuracy: 0.9350\n",
      "Epoch 51/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0785 - accuracy: 0.9864 - val_loss: 0.2402 - val_accuracy: 0.9300\n",
      "Epoch 52/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0812 - accuracy: 0.9838 - val_loss: 0.2420 - val_accuracy: 0.9300\n",
      "Epoch 53/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0762 - accuracy: 0.9850 - val_loss: 0.2390 - val_accuracy: 0.9300\n",
      "Epoch 54/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0792 - accuracy: 0.9862 - val_loss: 0.2424 - val_accuracy: 0.9350\n",
      "Epoch 55/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0736 - accuracy: 0.9827 - val_loss: 0.2425 - val_accuracy: 0.9300\n",
      "Epoch 56/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0779 - accuracy: 0.9860 - val_loss: 0.2411 - val_accuracy: 0.9300\n",
      "Epoch 57/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0726 - accuracy: 0.9833 - val_loss: 0.2411 - val_accuracy: 0.9350\n",
      "Epoch 58/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0624 - accuracy: 0.9878 - val_loss: 0.2424 - val_accuracy: 0.9350\n",
      "Epoch 59/150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0714 - accuracy: 0.9880 - val_loss: 0.2403 - val_accuracy: 0.9350\n",
      "Epoch 60/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0702 - accuracy: 0.9859 - val_loss: 0.2400 - val_accuracy: 0.9250\n",
      "Epoch 61/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0599 - accuracy: 0.9897 - val_loss: 0.2452 - val_accuracy: 0.9250\n",
      "Epoch 62/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0730 - accuracy: 0.9835 - val_loss: 0.2396 - val_accuracy: 0.9300\n",
      "Epoch 63/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0544 - accuracy: 0.9917 - val_loss: 0.2423 - val_accuracy: 0.9350\n",
      "Epoch 64/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0579 - accuracy: 0.9906 - val_loss: 0.2456 - val_accuracy: 0.9300\n",
      "Epoch 65/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0553 - accuracy: 0.9933 - val_loss: 0.2411 - val_accuracy: 0.9300\n",
      "Epoch 66/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0533 - accuracy: 0.9898 - val_loss: 0.2490 - val_accuracy: 0.9350\n",
      "Epoch 67/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0567 - accuracy: 0.9894 - val_loss: 0.2476 - val_accuracy: 0.9300\n",
      "Epoch 68/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0587 - accuracy: 0.9883 - val_loss: 0.2408 - val_accuracy: 0.9300\n",
      "Epoch 69/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0566 - accuracy: 0.9898 - val_loss: 0.2490 - val_accuracy: 0.9250\n",
      "Epoch 70/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0487 - accuracy: 0.9927 - val_loss: 0.2477 - val_accuracy: 0.9250\n",
      "Epoch 71/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0478 - accuracy: 0.9931 - val_loss: 0.2415 - val_accuracy: 0.9300\n",
      "Epoch 72/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0503 - accuracy: 0.9905 - val_loss: 0.2506 - val_accuracy: 0.9250\n",
      "Epoch 73/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0440 - accuracy: 0.9934 - val_loss: 0.2434 - val_accuracy: 0.9350\n",
      "Epoch 74/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0527 - accuracy: 0.9941 - val_loss: 0.2424 - val_accuracy: 0.9300\n",
      "Epoch 75/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0367 - accuracy: 0.9947 - val_loss: 0.2494 - val_accuracy: 0.9250\n",
      "Epoch 76/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0403 - accuracy: 0.9963 - val_loss: 0.2486 - val_accuracy: 0.9350\n",
      "Epoch 77/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0400 - accuracy: 0.9964 - val_loss: 0.2463 - val_accuracy: 0.9300\n",
      "Epoch 78/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0407 - accuracy: 0.9926 - val_loss: 0.2498 - val_accuracy: 0.9300\n",
      "Epoch 79/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0389 - accuracy: 0.9942 - val_loss: 0.2526 - val_accuracy: 0.9300\n",
      "Epoch 80/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0400 - accuracy: 0.9920 - val_loss: 0.2498 - val_accuracy: 0.9300\n",
      "Epoch 81/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0317 - accuracy: 0.9973 - val_loss: 0.2528 - val_accuracy: 0.9300\n",
      "Epoch 82/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0312 - accuracy: 0.9958 - val_loss: 0.2522 - val_accuracy: 0.9250\n",
      "Epoch 83/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0346 - accuracy: 0.9970 - val_loss: 0.2527 - val_accuracy: 0.9300\n",
      "Epoch 84/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0327 - accuracy: 0.9947 - val_loss: 0.2559 - val_accuracy: 0.9300\n",
      "Epoch 85/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0301 - accuracy: 0.9978 - val_loss: 0.2560 - val_accuracy: 0.9250\n",
      "Epoch 86/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0346 - accuracy: 0.9972 - val_loss: 0.2504 - val_accuracy: 0.9300\n",
      "Epoch 87/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0328 - accuracy: 0.9975 - val_loss: 0.2566 - val_accuracy: 0.9300\n",
      "Epoch 88/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0296 - accuracy: 0.9981 - val_loss: 0.2565 - val_accuracy: 0.9250\n",
      "Epoch 89/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0325 - accuracy: 0.9967 - val_loss: 0.2595 - val_accuracy: 0.9200\n",
      "Epoch 90/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0311 - accuracy: 0.9969 - val_loss: 0.2559 - val_accuracy: 0.9300\n",
      "Epoch 91/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0289 - accuracy: 0.9975 - val_loss: 0.2592 - val_accuracy: 0.9300\n",
      "Epoch 92/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0291 - accuracy: 0.9971 - val_loss: 0.2577 - val_accuracy: 0.9300\n",
      "Epoch 93/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0256 - accuracy: 0.9997 - val_loss: 0.2598 - val_accuracy: 0.9250\n",
      "Epoch 94/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0271 - accuracy: 0.9973 - val_loss: 0.2548 - val_accuracy: 0.9350\n",
      "Epoch 95/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0241 - accuracy: 0.9986 - val_loss: 0.2618 - val_accuracy: 0.9250\n",
      "Epoch 96/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0247 - accuracy: 0.9982 - val_loss: 0.2593 - val_accuracy: 0.9300\n",
      "Epoch 97/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0262 - accuracy: 0.9983 - val_loss: 0.2602 - val_accuracy: 0.9250\n",
      "Epoch 98/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0225 - accuracy: 0.9993 - val_loss: 0.2636 - val_accuracy: 0.9250\n",
      "Epoch 99/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0281 - accuracy: 0.9951 - val_loss: 0.2655 - val_accuracy: 0.9300\n",
      "Epoch 100/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0280 - accuracy: 0.9939 - val_loss: 0.2619 - val_accuracy: 0.9300\n",
      "Epoch 101/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0214 - accuracy: 0.9995 - val_loss: 0.2658 - val_accuracy: 0.9250\n",
      "Epoch 102/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0229 - accuracy: 0.9977 - val_loss: 0.2656 - val_accuracy: 0.9250\n",
      "Epoch 103/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0198 - accuracy: 0.9993 - val_loss: 0.2636 - val_accuracy: 0.9300\n",
      "Epoch 104/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0187 - accuracy: 0.9997 - val_loss: 0.2617 - val_accuracy: 0.9300\n",
      "Epoch 105/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0203 - accuracy: 0.9985 - val_loss: 0.2672 - val_accuracy: 0.9250\n",
      "Epoch 106/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0204 - accuracy: 0.9988 - val_loss: 0.2709 - val_accuracy: 0.9200\n",
      "Epoch 107/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0197 - accuracy: 0.9993 - val_loss: 0.2662 - val_accuracy: 0.9300\n",
      "Epoch 108/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0174 - accuracy: 0.9988 - val_loss: 0.2663 - val_accuracy: 0.9250\n",
      "Epoch 109/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0189 - accuracy: 0.9987 - val_loss: 0.2668 - val_accuracy: 0.9300\n",
      "Epoch 110/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.0168 - accuracy: 0.9988 - val_loss: 0.2691 - val_accuracy: 0.9300\n",
      "Epoch 111/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0174 - accuracy: 0.9991 - val_loss: 0.2676 - val_accuracy: 0.9300\n",
      "Epoch 112/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0176 - accuracy: 0.9995 - val_loss: 0.2726 - val_accuracy: 0.9250\n",
      "Epoch 113/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0174 - accuracy: 0.9993 - val_loss: 0.2692 - val_accuracy: 0.9250\n",
      "Epoch 114/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.0164 - accuracy: 0.9975 - val_loss: 0.2712 - val_accuracy: 0.9250\n",
      "Epoch 115/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0173 - accuracy: 0.9969 - val_loss: 0.2711 - val_accuracy: 0.9300\n",
      "Epoch 116/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0178 - accuracy: 0.9975 - val_loss: 0.2735 - val_accuracy: 0.9250\n",
      "Epoch 117/150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0147 - accuracy: 0.9991 - val_loss: 0.2711 - val_accuracy: 0.9200\n",
      "Epoch 118/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0163 - accuracy: 0.9984 - val_loss: 0.2719 - val_accuracy: 0.9250\n",
      "Epoch 119/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0146 - accuracy: 1.0000 - val_loss: 0.2746 - val_accuracy: 0.9200\n",
      "Epoch 120/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0144 - accuracy: 0.9997 - val_loss: 0.2726 - val_accuracy: 0.9300\n",
      "Epoch 121/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0137 - accuracy: 1.0000 - val_loss: 0.2747 - val_accuracy: 0.9200\n",
      "Epoch 122/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0131 - accuracy: 0.9998 - val_loss: 0.2765 - val_accuracy: 0.9300\n",
      "Epoch 123/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0120 - accuracy: 0.9995 - val_loss: 0.2750 - val_accuracy: 0.9250\n",
      "Epoch 124/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0130 - accuracy: 1.0000 - val_loss: 0.2743 - val_accuracy: 0.9250\n",
      "Epoch 125/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0132 - accuracy: 1.0000 - val_loss: 0.2779 - val_accuracy: 0.9200\n",
      "Epoch 126/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0118 - accuracy: 1.0000 - val_loss: 0.2758 - val_accuracy: 0.9300\n",
      "Epoch 127/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.0112 - accuracy: 1.0000 - val_loss: 0.2813 - val_accuracy: 0.9200\n",
      "Epoch 128/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0131 - accuracy: 0.9993 - val_loss: 0.2780 - val_accuracy: 0.9250\n",
      "Epoch 129/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0128 - accuracy: 1.0000 - val_loss: 0.2777 - val_accuracy: 0.9300\n",
      "Epoch 130/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0120 - accuracy: 1.0000 - val_loss: 0.2818 - val_accuracy: 0.9200\n",
      "Epoch 131/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0112 - accuracy: 1.0000 - val_loss: 0.2815 - val_accuracy: 0.9250\n",
      "Epoch 132/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0095 - accuracy: 1.0000 - val_loss: 0.2784 - val_accuracy: 0.9300\n",
      "Epoch 133/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0103 - accuracy: 1.0000 - val_loss: 0.2798 - val_accuracy: 0.9250\n",
      "Epoch 134/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0102 - accuracy: 1.0000 - val_loss: 0.2830 - val_accuracy: 0.9200\n",
      "Epoch 135/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0124 - accuracy: 1.0000 - val_loss: 0.2865 - val_accuracy: 0.9250\n",
      "Epoch 136/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0103 - accuracy: 0.9995 - val_loss: 0.2817 - val_accuracy: 0.9250\n",
      "Epoch 137/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0111 - accuracy: 1.0000 - val_loss: 0.2848 - val_accuracy: 0.9250\n",
      "Epoch 138/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0110 - accuracy: 0.9983 - val_loss: 0.2849 - val_accuracy: 0.9200\n",
      "Epoch 139/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0118 - accuracy: 1.0000 - val_loss: 0.2844 - val_accuracy: 0.9300\n",
      "Epoch 140/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.0084 - accuracy: 1.0000 - val_loss: 0.2886 - val_accuracy: 0.9200\n",
      "Epoch 141/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0111 - accuracy: 1.0000 - val_loss: 0.2870 - val_accuracy: 0.9250\n",
      "Epoch 142/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0099 - accuracy: 0.9995 - val_loss: 0.2876 - val_accuracy: 0.9200\n",
      "Epoch 143/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0105 - accuracy: 1.0000 - val_loss: 0.2890 - val_accuracy: 0.9250\n",
      "Epoch 144/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0076 - accuracy: 1.0000 - val_loss: 0.2894 - val_accuracy: 0.9250\n",
      "Epoch 145/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0095 - accuracy: 1.0000 - val_loss: 0.2869 - val_accuracy: 0.9250\n",
      "Epoch 146/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0081 - accuracy: 1.0000 - val_loss: 0.2914 - val_accuracy: 0.9200\n",
      "Epoch 147/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0091 - accuracy: 1.0000 - val_loss: 0.2898 - val_accuracy: 0.9250\n",
      "Epoch 148/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0076 - accuracy: 1.0000 - val_loss: 0.2931 - val_accuracy: 0.9200\n",
      "Epoch 149/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0078 - accuracy: 1.0000 - val_loss: 0.2919 - val_accuracy: 0.9200\n",
      "Epoch 150/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.0070 - accuracy: 1.0000 - val_loss: 0.2906 - val_accuracy: 0.9200\n"
     ]
    }
   ],
   "source": [
    "# 1. Import libraries and modules\n",
    "from tensorflow import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense\n",
    "import numpy as np\n",
    "from dataset import load_hoda\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline \n",
    "\n",
    "np.random.seed(123)  # for reproducibility\n",
    "\n",
    "# 2. Load pre-shuffled HODA data into train and test sets\n",
    "x_train, y_train_original, x_test, y_test_original = load_hoda()\n",
    "\n",
    "# 3. Preprocess input data\n",
    "'''3.2 normalize our data values to the range [0, 1]'''\n",
    "x_train = x_train.astype('float32')\n",
    "x_test = x_test.astype('float32')\n",
    "x_train /= 255\n",
    "x_test /= 255\n",
    "\n",
    "# 4. Preprocess class labels\n",
    "y_train = keras.utils.to_categorical(y_train_original, num_classes=10)\n",
    "y_test = keras.utils.to_categorical(y_test_original, num_classes=10)\n",
    "\n",
    "# 5. Define model architecture\n",
    "model = Sequential()\n",
    "model.add(Dense(60, activation='relu', input_dim=25))\n",
    "model.add(Dense(300, activation='relu'))\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "# 6. Compile model\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "\n",
    "history = model.fit(x_train, y_train,\n",
    "          epochs=150,\n",
    "          batch_size=128, validation_data=(x_test, y_test))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy: 0.9200000166893005 \n",
      "Maximum Test Acuracy During Training:  0.9350000023841858\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Test Accuracy:\",history.history['val_accuracy'][-1],\n",
    "      \"\\nMaximum Test Acuracy During Training: \", \n",
    "      max(history.history['val_accuracy']))\n",
    "\n",
    "plt.figure(1)\n",
    "plt.plot(history.history['loss'], label='Train')\n",
    "plt.plot(history.history['val_loss'], label='Test')\n",
    "plt.xlabel('#epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "plt.figure(2)\n",
    "plt.plot(history.history['accuracy'], label='Train')\n",
    "plt.plot(history.history['val_accuracy'], label='Test')\n",
    "plt.xlabel('#epochs')\n",
    "plt.ylabel('Accuracy(%)')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">شبکه با Dropout</div>\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/150\n",
      "8/8 [==============================] - 2s 64ms/step - loss: 2.2936 - accuracy: 0.1428 - val_loss: 2.1227 - val_accuracy: 0.4150\n",
      "Epoch 2/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 2.0810 - accuracy: 0.3200 - val_loss: 1.9229 - val_accuracy: 0.6500\n",
      "Epoch 3/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.8547 - accuracy: 0.5164 - val_loss: 1.6742 - val_accuracy: 0.7300\n",
      "Epoch 4/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.6218 - accuracy: 0.6114 - val_loss: 1.3801 - val_accuracy: 0.8000\n",
      "Epoch 5/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 1.3858 - accuracy: 0.6614 - val_loss: 1.1005 - val_accuracy: 0.8250\n",
      "Epoch 6/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 1.1382 - accuracy: 0.7186 - val_loss: 0.8825 - val_accuracy: 0.8500\n",
      "Epoch 7/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.9519 - accuracy: 0.7589 - val_loss: 0.7304 - val_accuracy: 0.8550\n",
      "Epoch 8/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.8503 - accuracy: 0.7495 - val_loss: 0.6205 - val_accuracy: 0.8550\n",
      "Epoch 9/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.7430 - accuracy: 0.7667 - val_loss: 0.5540 - val_accuracy: 0.8550\n",
      "Epoch 10/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.7045 - accuracy: 0.7787 - val_loss: 0.5052 - val_accuracy: 0.8550\n",
      "Epoch 11/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.6182 - accuracy: 0.8167 - val_loss: 0.4636 - val_accuracy: 0.8750\n",
      "Epoch 12/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.5528 - accuracy: 0.8218 - val_loss: 0.4327 - val_accuracy: 0.8750\n",
      "Epoch 13/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.5656 - accuracy: 0.8217 - val_loss: 0.4159 - val_accuracy: 0.8750\n",
      "Epoch 14/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.5312 - accuracy: 0.8170 - val_loss: 0.3925 - val_accuracy: 0.8750\n",
      "Epoch 15/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.4847 - accuracy: 0.8439 - val_loss: 0.3773 - val_accuracy: 0.8950\n",
      "Epoch 16/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.4832 - accuracy: 0.8488 - val_loss: 0.3626 - val_accuracy: 0.8950\n",
      "Epoch 17/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.4640 - accuracy: 0.8467 - val_loss: 0.3480 - val_accuracy: 0.9000\n",
      "Epoch 18/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.4316 - accuracy: 0.8620 - val_loss: 0.3413 - val_accuracy: 0.9050\n",
      "Epoch 19/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.4274 - accuracy: 0.8660 - val_loss: 0.3298 - val_accuracy: 0.9000\n",
      "Epoch 20/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.4364 - accuracy: 0.8625 - val_loss: 0.3255 - val_accuracy: 0.9050\n",
      "Epoch 21/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.4587 - accuracy: 0.8384 - val_loss: 0.3145 - val_accuracy: 0.9150\n",
      "Epoch 22/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.3619 - accuracy: 0.8862 - val_loss: 0.3089 - val_accuracy: 0.9100\n",
      "Epoch 23/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.4019 - accuracy: 0.8531 - val_loss: 0.3008 - val_accuracy: 0.9050\n",
      "Epoch 24/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3645 - accuracy: 0.8848 - val_loss: 0.2969 - val_accuracy: 0.9050\n",
      "Epoch 25/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3799 - accuracy: 0.8672 - val_loss: 0.2909 - val_accuracy: 0.9150\n",
      "Epoch 26/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.3754 - accuracy: 0.8729 - val_loss: 0.2874 - val_accuracy: 0.9150\n",
      "Epoch 27/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3410 - accuracy: 0.8874 - val_loss: 0.2862 - val_accuracy: 0.9150\n",
      "Epoch 28/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.3891 - accuracy: 0.8711 - val_loss: 0.2815 - val_accuracy: 0.9150\n",
      "Epoch 29/150\n",
      "8/8 [==============================] - 0s 10ms/step - loss: 0.3568 - accuracy: 0.8900 - val_loss: 0.2787 - val_accuracy: 0.9150\n",
      "Epoch 30/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3121 - accuracy: 0.8926 - val_loss: 0.2703 - val_accuracy: 0.9150\n",
      "Epoch 31/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3112 - accuracy: 0.8900 - val_loss: 0.2697 - val_accuracy: 0.9100\n",
      "Epoch 32/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3475 - accuracy: 0.8852 - val_loss: 0.2693 - val_accuracy: 0.9100\n",
      "Epoch 33/150\n",
      "8/8 [==============================] - 0s 10ms/step - loss: 0.3369 - accuracy: 0.8914 - val_loss: 0.2668 - val_accuracy: 0.9100\n",
      "Epoch 34/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3082 - accuracy: 0.9019 - val_loss: 0.2661 - val_accuracy: 0.9250\n",
      "Epoch 35/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3033 - accuracy: 0.9027 - val_loss: 0.2632 - val_accuracy: 0.9250\n",
      "Epoch 36/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.3185 - accuracy: 0.8939 - val_loss: 0.2593 - val_accuracy: 0.9250\n",
      "Epoch 37/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2990 - accuracy: 0.8990 - val_loss: 0.2606 - val_accuracy: 0.9250\n",
      "Epoch 38/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2922 - accuracy: 0.9216 - val_loss: 0.2578 - val_accuracy: 0.9150\n",
      "Epoch 39/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2711 - accuracy: 0.9077 - val_loss: 0.2555 - val_accuracy: 0.9200\n",
      "Epoch 40/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2914 - accuracy: 0.8950 - val_loss: 0.2561 - val_accuracy: 0.9150\n",
      "Epoch 41/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.3273 - accuracy: 0.8803 - val_loss: 0.2575 - val_accuracy: 0.9200\n",
      "Epoch 42/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2859 - accuracy: 0.8884 - val_loss: 0.2513 - val_accuracy: 0.9350\n",
      "Epoch 43/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2850 - accuracy: 0.8989 - val_loss: 0.2490 - val_accuracy: 0.9400\n",
      "Epoch 44/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2660 - accuracy: 0.9179 - val_loss: 0.2489 - val_accuracy: 0.9250\n",
      "Epoch 45/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2648 - accuracy: 0.9149 - val_loss: 0.2541 - val_accuracy: 0.9250\n",
      "Epoch 46/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2670 - accuracy: 0.9110 - val_loss: 0.2476 - val_accuracy: 0.9400\n",
      "Epoch 47/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2784 - accuracy: 0.9111 - val_loss: 0.2451 - val_accuracy: 0.9350\n",
      "Epoch 48/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2446 - accuracy: 0.9242 - val_loss: 0.2428 - val_accuracy: 0.9350\n",
      "Epoch 49/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2492 - accuracy: 0.9221 - val_loss: 0.2427 - val_accuracy: 0.9350\n",
      "Epoch 50/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2565 - accuracy: 0.9171 - val_loss: 0.2424 - val_accuracy: 0.9250\n",
      "Epoch 51/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2840 - accuracy: 0.9052 - val_loss: 0.2379 - val_accuracy: 0.9300\n",
      "Epoch 52/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2629 - accuracy: 0.9039 - val_loss: 0.2353 - val_accuracy: 0.9350\n",
      "Epoch 53/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2526 - accuracy: 0.9108 - val_loss: 0.2346 - val_accuracy: 0.9400\n",
      "Epoch 54/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2363 - accuracy: 0.9144 - val_loss: 0.2344 - val_accuracy: 0.9350\n",
      "Epoch 55/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2488 - accuracy: 0.9095 - val_loss: 0.2336 - val_accuracy: 0.9250\n",
      "Epoch 56/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2500 - accuracy: 0.9042 - val_loss: 0.2355 - val_accuracy: 0.9250\n",
      "Epoch 57/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2677 - accuracy: 0.8995 - val_loss: 0.2321 - val_accuracy: 0.9300\n",
      "Epoch 58/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2694 - accuracy: 0.9074 - val_loss: 0.2291 - val_accuracy: 0.9250\n",
      "Epoch 59/150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2578 - accuracy: 0.9071 - val_loss: 0.2297 - val_accuracy: 0.9350\n",
      "Epoch 60/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2282 - accuracy: 0.9219 - val_loss: 0.2292 - val_accuracy: 0.9250\n",
      "Epoch 61/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.2172 - accuracy: 0.9330 - val_loss: 0.2268 - val_accuracy: 0.9250\n",
      "Epoch 62/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.2611 - accuracy: 0.9103 - val_loss: 0.2255 - val_accuracy: 0.9300\n",
      "Epoch 63/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1922 - accuracy: 0.9407 - val_loss: 0.2199 - val_accuracy: 0.9350\n",
      "Epoch 64/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2418 - accuracy: 0.9039 - val_loss: 0.2160 - val_accuracy: 0.9400\n",
      "Epoch 65/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2329 - accuracy: 0.9105 - val_loss: 0.2200 - val_accuracy: 0.9300\n",
      "Epoch 66/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2040 - accuracy: 0.9302 - val_loss: 0.2271 - val_accuracy: 0.9150\n",
      "Epoch 67/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2401 - accuracy: 0.9230 - val_loss: 0.2227 - val_accuracy: 0.9250\n",
      "Epoch 68/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2173 - accuracy: 0.9334 - val_loss: 0.2220 - val_accuracy: 0.9250\n",
      "Epoch 69/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2037 - accuracy: 0.9289 - val_loss: 0.2185 - val_accuracy: 0.9250\n",
      "Epoch 70/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2089 - accuracy: 0.9244 - val_loss: 0.2167 - val_accuracy: 0.9300\n",
      "Epoch 71/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2341 - accuracy: 0.9184 - val_loss: 0.2238 - val_accuracy: 0.9400\n",
      "Epoch 72/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2032 - accuracy: 0.9378 - val_loss: 0.2257 - val_accuracy: 0.9300\n",
      "Epoch 73/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2167 - accuracy: 0.9346 - val_loss: 0.2141 - val_accuracy: 0.9300\n",
      "Epoch 74/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2046 - accuracy: 0.9169 - val_loss: 0.2134 - val_accuracy: 0.9300\n",
      "Epoch 75/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.2331 - accuracy: 0.9163 - val_loss: 0.2124 - val_accuracy: 0.9300\n",
      "Epoch 76/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2074 - accuracy: 0.9320 - val_loss: 0.2155 - val_accuracy: 0.9250\n",
      "Epoch 77/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1827 - accuracy: 0.9358 - val_loss: 0.2132 - val_accuracy: 0.9250\n",
      "Epoch 78/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.2003 - accuracy: 0.9314 - val_loss: 0.2135 - val_accuracy: 0.9300\n",
      "Epoch 79/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1903 - accuracy: 0.9372 - val_loss: 0.2135 - val_accuracy: 0.9300\n",
      "Epoch 80/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.2021 - accuracy: 0.9371 - val_loss: 0.2128 - val_accuracy: 0.9350\n",
      "Epoch 81/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1840 - accuracy: 0.9286 - val_loss: 0.2194 - val_accuracy: 0.9250\n",
      "Epoch 82/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1922 - accuracy: 0.9245 - val_loss: 0.2179 - val_accuracy: 0.9250\n",
      "Epoch 83/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1993 - accuracy: 0.9312 - val_loss: 0.2127 - val_accuracy: 0.9250\n",
      "Epoch 84/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1855 - accuracy: 0.9315 - val_loss: 0.2154 - val_accuracy: 0.9250\n",
      "Epoch 85/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1875 - accuracy: 0.9256 - val_loss: 0.2188 - val_accuracy: 0.9250\n",
      "Epoch 86/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1844 - accuracy: 0.9327 - val_loss: 0.2173 - val_accuracy: 0.9300\n",
      "Epoch 87/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.2043 - accuracy: 0.9372 - val_loss: 0.2188 - val_accuracy: 0.9200\n",
      "Epoch 88/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1967 - accuracy: 0.9310 - val_loss: 0.2187 - val_accuracy: 0.9200\n",
      "Epoch 89/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1819 - accuracy: 0.9357 - val_loss: 0.2123 - val_accuracy: 0.9250\n",
      "Epoch 90/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1691 - accuracy: 0.9384 - val_loss: 0.2175 - val_accuracy: 0.9250\n",
      "Epoch 91/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1793 - accuracy: 0.9401 - val_loss: 0.2229 - val_accuracy: 0.9300\n",
      "Epoch 92/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1794 - accuracy: 0.9315 - val_loss: 0.2208 - val_accuracy: 0.9200\n",
      "Epoch 93/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1411 - accuracy: 0.9467 - val_loss: 0.2108 - val_accuracy: 0.9250\n",
      "Epoch 94/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1909 - accuracy: 0.9323 - val_loss: 0.2160 - val_accuracy: 0.9250\n",
      "Epoch 95/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1581 - accuracy: 0.9490 - val_loss: 0.2223 - val_accuracy: 0.9200\n",
      "Epoch 96/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1488 - accuracy: 0.9528 - val_loss: 0.2203 - val_accuracy: 0.9250\n",
      "Epoch 97/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1645 - accuracy: 0.9505 - val_loss: 0.2208 - val_accuracy: 0.9300\n",
      "Epoch 98/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1696 - accuracy: 0.9389 - val_loss: 0.2163 - val_accuracy: 0.9300\n",
      "Epoch 99/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1500 - accuracy: 0.9519 - val_loss: 0.2143 - val_accuracy: 0.9350\n",
      "Epoch 100/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1741 - accuracy: 0.9410 - val_loss: 0.2168 - val_accuracy: 0.9350\n",
      "Epoch 101/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1676 - accuracy: 0.9375 - val_loss: 0.2174 - val_accuracy: 0.9300\n",
      "Epoch 102/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1523 - accuracy: 0.9437 - val_loss: 0.2202 - val_accuracy: 0.9300\n",
      "Epoch 103/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1427 - accuracy: 0.9559 - val_loss: 0.2166 - val_accuracy: 0.9300\n",
      "Epoch 104/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1805 - accuracy: 0.9289 - val_loss: 0.2124 - val_accuracy: 0.9400\n",
      "Epoch 105/150\n",
      "8/8 [==============================] - 0s 8ms/step - loss: 0.1883 - accuracy: 0.9334 - val_loss: 0.2132 - val_accuracy: 0.9300\n",
      "Epoch 106/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1375 - accuracy: 0.9537 - val_loss: 0.2197 - val_accuracy: 0.9300\n",
      "Epoch 107/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1621 - accuracy: 0.9396 - val_loss: 0.2224 - val_accuracy: 0.9300\n",
      "Epoch 108/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1396 - accuracy: 0.9405 - val_loss: 0.2233 - val_accuracy: 0.9300\n",
      "Epoch 109/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1447 - accuracy: 0.9486 - val_loss: 0.2227 - val_accuracy: 0.9250\n",
      "Epoch 110/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1427 - accuracy: 0.9508 - val_loss: 0.2217 - val_accuracy: 0.9300\n",
      "Epoch 111/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1360 - accuracy: 0.9543 - val_loss: 0.2225 - val_accuracy: 0.9350\n",
      "Epoch 112/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1392 - accuracy: 0.9505 - val_loss: 0.2159 - val_accuracy: 0.9300\n",
      "Epoch 113/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1650 - accuracy: 0.9382 - val_loss: 0.2218 - val_accuracy: 0.9350\n",
      "Epoch 114/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.1693 - accuracy: 0.9417 - val_loss: 0.2251 - val_accuracy: 0.9300\n",
      "Epoch 115/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1464 - accuracy: 0.9463 - val_loss: 0.2231 - val_accuracy: 0.9250\n",
      "Epoch 116/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1309 - accuracy: 0.9544 - val_loss: 0.2151 - val_accuracy: 0.9300\n",
      "Epoch 117/150\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1268 - accuracy: 0.9492 - val_loss: 0.2184 - val_accuracy: 0.9400\n",
      "Epoch 118/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1279 - accuracy: 0.9583 - val_loss: 0.2219 - val_accuracy: 0.9400\n",
      "Epoch 119/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1239 - accuracy: 0.9597 - val_loss: 0.2306 - val_accuracy: 0.9300\n",
      "Epoch 120/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1312 - accuracy: 0.9594 - val_loss: 0.2301 - val_accuracy: 0.9300\n",
      "Epoch 121/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1319 - accuracy: 0.9616 - val_loss: 0.2313 - val_accuracy: 0.9250\n",
      "Epoch 122/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1403 - accuracy: 0.9564 - val_loss: 0.2318 - val_accuracy: 0.9250\n",
      "Epoch 123/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1417 - accuracy: 0.9535 - val_loss: 0.2223 - val_accuracy: 0.9300\n",
      "Epoch 124/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1272 - accuracy: 0.9598 - val_loss: 0.2185 - val_accuracy: 0.9350\n",
      "Epoch 125/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1211 - accuracy: 0.9581 - val_loss: 0.2221 - val_accuracy: 0.9350\n",
      "Epoch 126/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1289 - accuracy: 0.9613 - val_loss: 0.2240 - val_accuracy: 0.9300\n",
      "Epoch 127/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1167 - accuracy: 0.9611 - val_loss: 0.2216 - val_accuracy: 0.9300\n",
      "Epoch 128/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1127 - accuracy: 0.9594 - val_loss: 0.2188 - val_accuracy: 0.9400\n",
      "Epoch 129/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1089 - accuracy: 0.9629 - val_loss: 0.2175 - val_accuracy: 0.9350\n",
      "Epoch 130/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1339 - accuracy: 0.9442 - val_loss: 0.2243 - val_accuracy: 0.9300\n",
      "Epoch 131/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1586 - accuracy: 0.9428 - val_loss: 0.2317 - val_accuracy: 0.9250\n",
      "Epoch 132/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1167 - accuracy: 0.9623 - val_loss: 0.2299 - val_accuracy: 0.9350\n",
      "Epoch 133/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1082 - accuracy: 0.9695 - val_loss: 0.2310 - val_accuracy: 0.9300\n",
      "Epoch 134/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1299 - accuracy: 0.9657 - val_loss: 0.2272 - val_accuracy: 0.9350\n",
      "Epoch 135/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1308 - accuracy: 0.9621 - val_loss: 0.2262 - val_accuracy: 0.9350\n",
      "Epoch 136/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1198 - accuracy: 0.9506 - val_loss: 0.2195 - val_accuracy: 0.9350\n",
      "Epoch 137/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1107 - accuracy: 0.9668 - val_loss: 0.2200 - val_accuracy: 0.9450\n",
      "Epoch 138/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1636 - accuracy: 0.9316 - val_loss: 0.2283 - val_accuracy: 0.9300\n",
      "Epoch 139/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1196 - accuracy: 0.9602 - val_loss: 0.2403 - val_accuracy: 0.9200\n",
      "Epoch 140/150\n",
      "8/8 [==============================] - 0s 5ms/step - loss: 0.1334 - accuracy: 0.9536 - val_loss: 0.2359 - val_accuracy: 0.9300\n",
      "Epoch 141/150\n",
      "8/8 [==============================] - 0s 9ms/step - loss: 0.0977 - accuracy: 0.9703 - val_loss: 0.2382 - val_accuracy: 0.9300\n",
      "Epoch 142/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1093 - accuracy: 0.9586 - val_loss: 0.2373 - val_accuracy: 0.9250\n",
      "Epoch 143/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1132 - accuracy: 0.9632 - val_loss: 0.2297 - val_accuracy: 0.9350\n",
      "Epoch 144/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1117 - accuracy: 0.9520 - val_loss: 0.2313 - val_accuracy: 0.9350\n",
      "Epoch 145/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.0980 - accuracy: 0.9653 - val_loss: 0.2288 - val_accuracy: 0.9350\n",
      "Epoch 146/150\n",
      "8/8 [==============================] - 0s 6ms/step - loss: 0.1108 - accuracy: 0.9649 - val_loss: 0.2332 - val_accuracy: 0.9400\n",
      "Epoch 147/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1120 - accuracy: 0.9581 - val_loss: 0.2355 - val_accuracy: 0.9400\n",
      "Epoch 148/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1223 - accuracy: 0.9536 - val_loss: 0.2286 - val_accuracy: 0.9250\n",
      "Epoch 149/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1002 - accuracy: 0.9681 - val_loss: 0.2221 - val_accuracy: 0.9300\n",
      "Epoch 150/150\n",
      "8/8 [==============================] - 0s 7ms/step - loss: 0.1103 - accuracy: 0.9661 - val_loss: 0.2211 - val_accuracy: 0.9350\n"
     ]
    }
   ],
   "source": [
    "# 1. Import libraries and modules\n",
    "from tensorflow import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout\n",
    "import numpy as np\n",
    "from dataset import load_hoda\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline \n",
    "\n",
    "np.random.seed(123)  # for reproducibility\n",
    "\n",
    "# 2. Load pre-shuffled HODA data into train and test sets\n",
    "x_train, y_train_original, x_test, y_test_original = load_hoda()\n",
    "\n",
    "# 3. Preprocess input data\n",
    "'''3.2 normalize our data values to the range [0, 1]'''\n",
    "x_train = x_train.astype('float32')\n",
    "x_test = x_test.astype('float32')\n",
    "x_train /= 255\n",
    "x_test /= 255\n",
    "\n",
    "# 4. Preprocess class labels\n",
    "y_train = keras.utils.to_categorical(y_train_original, num_classes=10)\n",
    "y_test = keras.utils.to_categorical(y_test_original, num_classes=10)\n",
    "\n",
    "# 5. Define model architecture\n",
    "model = Sequential()\n",
    "model.add(Dense(60, activation='relu', input_dim=25))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(300, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(10, activation='softmax'))\n",
    "\n",
    "# 6. Compile model\n",
    "model.compile(loss='categorical_crossentropy',\n",
    "              optimizer='adam',\n",
    "              metrics=['accuracy'])\n",
    "\n",
    "\n",
    "history = model.fit(x_train, y_train,\n",
    "          epochs=150,\n",
    "          batch_size=128, validation_data=(x_test, y_test))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test Accuracy: 0.9350000023841858 \n",
      "Maximum Test Acuracy During Training:  0.9449999928474426\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "print(\"Test Accuracy:\",history.history['val_accuracy'][-1],\n",
    "      \"\\nMaximum Test Acuracy During Training: \", \n",
    "      max(history.history['val_accuracy']))\n",
    "\n",
    "plt.figure(1)\n",
    "plt.plot(history.history['loss'], label='Train')\n",
    "plt.plot(history.history['val_loss'], label='Test')\n",
    "plt.xlabel('#epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()\n",
    "\n",
    "plt.figure(2)\n",
    "plt.plot(history.history['accuracy'], label='Train')\n",
    "plt.plot(history.history['val_accuracy'], label='Test')\n",
    "plt.xlabel('#epochs')\n",
    "plt.ylabel('Accuracy(%)')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "tf2.5",
   "language": "python",
   "name": "tensorflow_25"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  },
  "nbpresent": {
   "slides": {
    "300ee14f-a043-486e-b274-7ff253907cd7": {
     "id": "300ee14f-a043-486e-b274-7ff253907cd7",
     "prev": "cb74e0bc-4513-4d13-b7f1-14c3078a7927",
     "regions": {
      "26dc3f39-a230-447c-af4c-f5e5b2fb7835": {
       "attrs": {
        "height": 0.8,
        "width": 0.8,
        "x": 0.1,
        "y": 0.1
       },
       "content": {
        "cell": "c58440a5-3f8f-4f37-9c79-6bf766209406",
        "part": "whole"
       },
       "id": "26dc3f39-a230-447c-af4c-f5e5b2fb7835"
      }
     }
    },
    "878aa53a-1444-4100-8f50-7a408191c579": {
     "id": "878aa53a-1444-4100-8f50-7a408191c579",
     "prev": null,
     "regions": {
      "a6c6843a-5ea6-4fbc-b890-3b4b8ae475b3": {
       "attrs": {
        "height": 0.8,
        "width": 0.8,
        "x": 0.1,
        "y": 0.1
       },
       "content": {
        "cell": "588ee1fa-64b5-453b-ade7-8e6b2515821c",
        "part": "whole"
       },
       "id": "a6c6843a-5ea6-4fbc-b890-3b4b8ae475b3"
      }
     }
    },
    "96ffe88e-7b50-43de-afdd-942e564f4e3e": {
     "id": "96ffe88e-7b50-43de-afdd-942e564f4e3e",
     "prev": "878aa53a-1444-4100-8f50-7a408191c579",
     "regions": {
      "b7e52e12-489a-468d-b10c-af2024fd2856": {
       "attrs": {
        "height": 0.8,
        "width": 0.8,
        "x": 0.1,
        "y": 0.1
       },
       "content": {
        "cell": "de829a92-1fb6-44ad-a2c6-fc1001e1f6e1",
        "part": "whole"
       },
       "id": "b7e52e12-489a-468d-b10c-af2024fd2856"
      }
     }
    },
    "cb74e0bc-4513-4d13-b7f1-14c3078a7927": {
     "id": "cb74e0bc-4513-4d13-b7f1-14c3078a7927",
     "prev": "96ffe88e-7b50-43de-afdd-942e564f4e3e",
     "regions": {
      "444878ee-68f3-4abb-acff-a7079b21e86d": {
       "attrs": {
        "height": 0.8,
        "width": 0.8,
        "x": 0.1,
        "y": 0.1
       },
       "content": {
        "cell": "25f3f538-1ee8-4d98-a6bb-14cbeb7a702d",
        "part": "whole"
       },
       "id": "444878ee-68f3-4abb-acff-a7079b21e86d"
      }
     }
    }
   },
   "themes": {}
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}