03 - Result processing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "import seaborn as sns\n",
    "colors = sns.color_palette()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(216, 11)\n",
      "['final_learning_rate' 'include_digit_length_classifier' 'layers'\n",
      " 'learning_rate' 'num_hidden' 'step' 'test_accuracy' 'train_accuracy'\n",
      " 'use_dropout' 'use_max_pool' 'valid_accuracy']\n"
     ]
    }
   ],
   "source": [
    "df_2 = pd.read_csv('results-2mins.bak.csv')\n",
    "df_5 = pd.read_csv('results-5mins.bak.csv')\n",
    "print(df_2.shape)\n",
    "print(df_2.columns.values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\navneet\\desktop\\udacity\\ml\\deeplearning\\digicam\\myvenv\\lib\\site-packages\\matplotlib\\axes\\_axes.py:6462: UserWarning: The 'normed' kwarg is deprecated, and has been replaced by the 'density' kwarg.\n",
      "  warnings.warn(\"The 'normed' kwarg is deprecated, and has been \"\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "count    216.000000\n",
      "mean       0.569963\n",
      "std        0.106044\n",
      "min        0.421000\n",
      "25%        0.470250\n",
      "50%        0.550500\n",
      "75%        0.670250\n",
      "max        0.798000\n",
      "Name: test_accuracy, dtype: float64\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAGzFJREFUeJzt3XuUVOWd7vHvw02IIiigqBgaJprhojbY4jVrAUnUEEWM8YImkeDRORijidGMeqJmnODgjCeaMc5xEeMtGsUBjYQTJyaAGoyJNtDiABoMgnb0IKKIEokgv/PH3kCBDVVI713dvZ/PWrXY9/rV2009/e5d9W5FBGZmVlztql2AmZlVl4PAzKzgHARmZgXnIDAzKzgHgZlZwTkIzMwKzkFguZH0fUn3Znj8hZKGp9OSdKektyU9I+kzkl7M6rnNWjMHgTUrSWdLqpf0nqTXJT0q6bg8njsiBkXE4+nsccDngT4RMSwifhcRn26u55J0l6QfNMNxaiSFpA7NUZfZx+EgsGYj6VLgZuB6YF/gk8B/AKdUoZy+wLKIWLurB2rrb9Jt/fVZeQ4CaxaSugHXAd+IiIciYm1ErI+IX0bE5dvZ5z8l/T9J70h6UtKgknWjJC2S9K6kv0i6LF3eU9IMSaslvSXpd5LapeuWSfqcpPOA24Gj057JP0kaLqmx5Pj7S5omaaWklyVdXLLu+5KmSrpX0hpg3DZ1XwCcA3w3Pf4vKzjmsLSntEbSCkk/TFc9mf67Oj3W0U200/b2RdJxkn6ftserksZt+nlIuietZbmk75W00zhJT0m6SdJbwPfT5eMlLU5Pp/1aUt90udJt30h/VgskDd7Or4K1RhHhhx+7/ABOBDYAHXawzfeBe0vmxwNdgd1IehINJeteBz6TTu8FDE2n/wW4DeiYPj4DKF23DPhcOj0OmFNyvOFAYzrdDpgLXAN0AvoDS4ETSupcD4xJt+3SxGu5C/hByXy5Yz4NfDWd3gM4Kp2uAaJMu21v308C7wJj07boAdSm6+4BHknbtwb4E3BeSdtsAL4JdAC6pK/1JWBAuux7wO/T7U9IX1t3QOk2+1X7d86P5nu4R2DNpQfwZkRsqHSHiLgjIt6NiL+RvPkelvYsIHkjHihpz4h4OyLmlSzfD+gbSY/jd5G+W+2EI4BeEXFdRHwQEUuBnwBnlWzzdET8IiI2RsT7zXDM9cCnJPWMiPci4g87Ue/29j0H+G1E3J+2xaqIaJDUHjgTuDJt32XA/wa+WnLM1yLilojYkL6+fwD+JSIWpz/D64HatFewniRQ/p4kdBdHxOs7Ub+1cA4Cay6rgJ6Vnm+W1F7SJEl/Tk+/LEtX9Uz/PQ0YBSyX9ETJKZN/I/nL9TFJSyVd8TFq7Qvsn55OWS1pNXAVyXWNTV5t5mOeBxwMvCDpWUkn7cSxt7fvgcCfm9i+J0mvZHnJsuXAASXz276+vsCPSmp/i+Sv/wMiYhbwY+BWYIWkyZL23In6rYVzEFhzeRpYR3KKoRJnk1xE/hzQjeT0BSRvPkTEsxFxCrAP8AvgwXT5uxHxnYjoD5wMXCrpsztZ66vAyxHRveTRNSJGlWxTrpex7fodHjMilkTE2PT13ABMlbR7Bc+zo31fBf6uiV3eJPkrvm/Jsk8CfylT/z9sU3+XiPh9WsO/R8ThwCCSUGryuo+1Tg4CaxYR8Q7J+fFbJY2R9AlJHSV9QdK/NrFLV+BvJD2JT5CcigBAUidJ50jqFhHrgTXAh+m6kyR9SpJKln+4k+U+A6yR9I+SuqS9k8GSjtiJY6wguQ5Q0TElfUVSr4jYCKxO9/kQWAls3OZYW9nBvvcBn5N0hqQOknpIqo2ID0mCc6KkrunpnUuBHX2H4zbgSqUX7NOLzaen00dIOlJSR2AtSeDvbJtbC+YgsGYTET8kecP5Hskb3KvARSR/0W/rHpLTFX8BFgHbnjP/KrAsPW30P4GvpMsPAn4LvEfSC/mP2PLdgUrr/JCkN1ELvEzyF/TtJD2TSv2U5BrGakm/qOCYJwILJb0H/Ag4KyLWRcRfgYnAU+mxjmriuba37yskp8++Q3IqpwE4LN3nmyRv2kuBOcDPgTt20CYPk/Q2Hkjb/L+BL6Sr9yS53vE2yc9sFXBj5U1lLd2mT1uYmVlBuUdgZlZwDgIzs4JzEJiZFZyDwMys4FrFYFM9e/aMmpqaapdhZtaqzJ07982I6FVuu1YRBDU1NdTX11e7DDOzVkXS8vJb+dSQmVnhOQjMzArOQWBmVnCt4hpBU9avX09jYyPr1q2rdimWg86dO9OnTx86duxY7VLM2pxWGwSNjY107dqVmpoakvHHrK2KCFatWkVjYyP9+vWrdjlmbU6rPTW0bt06evTo4RAoAEn06NHDvT+zjLTaIAAcAgXin7VZdlp1EJiZ2a5rtdcIPqL+zuY9Xt3Xy25SU1ND165dad++PR06dNilL72NGjWKn//853Tv3n2n921oaOC1115j1KhR5Tcu4/rrr+eqq67a5eNsMn36dBYtWsQVV3ycO0qaWR7aThBUyezZs+nZs2f5Dcv41a9+9bH3bWhooL6+PtMgiAgignbtdq4TOXr0aEaPHr3LdVkr1Nx/nFVbBX8ctlY+NZSxcePGMWHCBEaMGEH//v154oknGD9+PAMGDGDcuHGbt6upqeHNN99k2bJlDBgwgPPPP59BgwZx/PHH8/777wMwfPjwzb2ON998k5qaGj744AOuueYapkyZQm1tLVOmTGHt2rWMHz+eI444giFDhvDII48AsHDhQoYNG0ZtbS2HHnooS5Ys2arWK664gvfff5/a2lrOOeeczbVceOGFDB06lFdffZUJEyZQV1fHoEGDuPbaa7eq/9prr2Xo0KEccsghvPDCCwDcddddXHTRRZvb4uKLL+aYY46hf//+TJ06FYCNGzdy4YUXMmjQIE466SRGjRq1eZ2ZZc9BsAskcfzxx3P44YczefLk7W739ttvM2vWLG666SZOPvlkvv3tb7Nw4UKef/55GhoaPrL9kiVL+MY3vsHChQvp3r0706ZN2+6xO3XqxHXXXceZZ55JQ0MDZ555JhMnTmTkyJE8++yzzJ49m8svv5y1a9dy2223cckll2zuQfTp02erY02aNIkuXbrQ0NDAfffdB8CLL77I1772NebPn0/fvn2ZOHEi9fX1LFiwgCeeeIIFCxZs3r9nz57MmzePCRMmcOONTd/J8PXXX2fOnDnMmDFj8+mihx56iGXLlvH8889z++238/TTT2+/0c2s2TkIdsFTTz3FvHnzePTRR7n11lt58sknm9zu5JNPRhKHHHII++67L4cccgjt2rVj0KBBLFu27CPb9+vXj9raWgAOP/zwJrfZkccee4xJkyZRW1vL8OHDWbduHa+88gpHH300119/PTfccAPLly+nS5cuZY/Vt29fjjpqy210H3zwQYYOHcqQIUNYuHAhixYt2rzuS1/6Utmax4wZQ7t27Rg4cCArVqwAYM6cOZx++um0a9eO3r17M2LEiJ16vWa2axwEu2D//fcHYJ999uHUU0/lmWeeaXK73XbbDYB27dptnt40v2HDhu1uD9C+ffvN23To0IGNGzcC7PAz9RHBtGnTaGhooKGhgVdeeYUBAwZw9tlnM336dLp06cIJJ5zArFmzyr7G3XffffP0yy+/zI033sjMmTNZsGABX/ziF7eqY1PdpTXv6LVtul+275ttVl0Ogo9p7dq1vPvuu5unH3vsMQYPHpzpc9bU1DB37lyArc6hd+3adXMtACeccAK33HLL5jfY+fPnA7B06VL69+/PxRdfzOjRo7c6rbNJx44dWb9+fZPPv2bNGnbffXe6devGihUrePTRR5vldR133HFMmzaNjRs3smLFCh5//PFmOa6ZVabtfGoo5yv6K1as4NRTTwVgw4YNnH322Zx44omZPudll13GGWecwc9+9jNGjhy5efmIESM2nwq68sorufrqq/nWt77FoYceSkRQU1PDjBkzmDJlCvfeey8dO3akd+/eXHPNNR95jgsuuIBDDz2UoUOHMnHixK3WHXbYYQwZMoRBgwbRv39/jj322GZ5XaeddhozZ85k8ODBHHzwwRx55JF069atWY5tZuWpNXTL6+rqYtvP6C9evJgBAwZUqSJrbu+99x577LEHq1atYtiwYTz11FP07t17q238M29l/PHRqpM0NyLqym3XdnoE1qqddNJJrF69mg8++ICrr776IyFgZtlxEFiL4OsCZtXTqi8Wt4bTWtY8/LM2y06rDYLOnTuzatUqv0EUwKb7EXTu3LnapZi1Sa321FCfPn1obGxk5cqV1S7FcrDpDmVm1vxabRB07NjRd6syM2sGrfbUkJmZNQ8HgZlZwTkIzMwKzkFgZlZwDgIzs4JzEJiZFZyDwMys4BwEZmYF5yAwMys4B4GZWcE5CMzMCs5BYGZWcA4CM7OCyzwIJLWXNF/SjHS+n6Q/SloiaYqkTlnXYGZm25dHj+ASYHHJ/A3ATRFxEPA2cF4ONZiZ2XZkGgSS+gBfBG5P5wWMBKamm9wNjMmyBjMz27GsewQ3A98FNqbzPYDVEbEhnW8EDsi4BjMz24HMgkDSScAbETG3dHETmzZ502FJF0iql1Tv21GamWUnyx7BscBoScuAB0hOCd0MdJe06RaZfYDXmto5IiZHRF1E1PXq1SvDMs3Mii2zIIiIKyOiT0TUAGcBsyLiHGA28OV0s3OBR7KqwczMyqvG9wj+EbhU0ksk1wx+WoUazMws1aH8JrsuIh4HHk+nlwLD8nheMzMrz98sNjMrOAeBmVnBOQjMzArOQWBmVnAOAjOzgnMQmJkVnIPAzKzgHARmZgXnIDAzKzgHgZlZwTkIzMwKzkFgZlZwDgIzs4JzEJiZFZyDwMys4BwEZmYF5yAwMyu4XO5QZmYtSP2d1a7AWhj3CMzMCs5BYGZWcA4CM7OCcxCYmRWcg8DMrOAcBGZmBecgMDMrOAeBmVnBOQjMzArOQWBmVnAOAjOzgnMQmJkVnIPAzKzgHARmZgXnIDAzKzgHgZlZwTkIzMwKLrMgkNRZ0jOSnpO0UNI/pcv7SfqjpCWSpkjqlFUNZmZWXpY9gr8BIyPiMKAWOFHSUcANwE0RcRDwNnBehjWYmVkZmQVBJN5LZzumjwBGAlPT5XcDY7KqwczMysv0GoGk9pIagDeA3wB/BlZHxIZ0k0bggCxrMDOzHcs0CCLiw4ioBfoAw4ABTW3W1L6SLpBUL6l+5cqVWZZpZlZouXxqKCJWA48DRwHdJXVIV/UBXtvOPpMjoi4i6nr16pVHmWZmhZTlp4Z6SeqeTncBPgcsBmYDX043Oxd4JKsazMysvA7lN/nY9gPultSeJHAejIgZkhYBD0j6ATAf+GmGNZiZWRmZBUFELACGNLF8Kcn1AjOz1qP+znyep+7r+TxPCX+z2Mys4BwEZmYF5yAwMys4B4GZWcE5CMzMCs5BYGZWcBUFgaRjK1lmZmatT6U9glsqXGZmZq3MDr9QJulo4Bigl6RLS1btCbTPsjAzM8tHuW8WdwL2SLfrWrJ8DVvGCzIzs1Zsh0EQEU8AT0i6KyKW51STmZnlqNKxhnaTNBmoKd0nIkZmUZSZmeWn0iD4T+A24Hbgw+zKMTOzvFUaBBsi4v9kWomZmVVFpR8f/aWkCyXtJ2nvTY9MKzMzs1xU2iM4N/338pJlAfRv3nLMzCxvFQVBRPTLuhAzM6uOioJA0teaWh4R9zRvOWZmlrdKTw0dUTLdGfgsMA9wEJiZtXKVnhr6Zum8pG7AzzKpyMzMcvVxh6H+K3BQcxZiZmbVUek1gl+SfEoIksHmBgAPZlWUmZnlp9JrBDeWTG8AlkdEYwb1mJlZzio6NZQOPvcCyQikewEfZFmUmZnlp9I7lJ0BPAOcDpwB/FGSh6E2M2sDKj019L+AIyLiDQBJvYDfAlOzKszMzPJR6aeG2m0KgdSqndjXzMxasEp7BP8l6dfA/en8mcCvsinJzMzyVO6exZ8C9o2IyyV9CTgOEPA0cF8O9ZmZWcbKnd65GXgXICIeiohLI+LbJL2Bm7MuzszMslcuCGoiYsG2CyOinuS2lWZm1sqVC4LOO1jXpTkLMTOz6igXBM9KOn/bhZLOA+ZmU5KZmeWp3KeGvgU8LOkctrzx1wGdgFOzLMzMzPKxwyCIiBXAMZJGAIPTxf83ImZlXpmZmeWi0vsRzAZmZ1yLmZlVQWbfDpZ0oKTZkhZLWijpknT53pJ+I2lJ+u9eWdVgZmblZTlMxAbgOxExADgK+IakgcAVwMyIOAiYmc6bmVmVZBYEEfF6RMxLp98FFgMHAKcAd6eb3Q2MyaoGMzMrL5eB4yTVAEOAP5IMWfE6JGEB7LOdfS6QVC+pfuXKlXmUaWZWSJkHgaQ9gGnAtyJiTaX7RcTkiKiLiLpevXplV6CZWcFlGgSSOpKEwH0R8VC6eIWk/dL1+wFvbG9/MzPLXpafGhLwU2BxRPywZNV04Nx0+lzgkaxqMDOz8iq9H8HHcSzwVeB5SQ3psquAScCD6TAVr5Dc/tLMzKoksyCIiDkk9y5oymezel4zM9s5vt2kmVnBOQjMzAouy2sEZm1H/Z35PE/d1/N5HrMS7hGYmRWcg8DMrOAcBGZmBecgMDMrOF8sbil8MdLMqsQ9AjOzgnMQmJkVnIPAzKzgfI3ArCXJ61qRWQn3CMzMCs5BYGZWcA4CM7OCcxCYmRWcg8DMrOAcBGZmBecgMDMrOAeBmVnBOQjMzArOQWBmVnAOAjOzgnMQmJkVnIPAzKzgHARmZgXnIDAzKzgHgZlZwTkIzMwKzkFgZlZwDgIzs4JzEJiZFZyDwMys4BwEZmYF5yAwMyu4zIJA0h2S3pD03yXL9pb0G0lL0n/3yur5zcysMln2CO4CTtxm2RXAzIg4CJiZzpuZWRVlFgQR8STw1jaLTwHuTqfvBsZk9fxmZlaZvK8R7BsRrwOk/+6zvQ0lXSCpXlL9ypUrcyvQzKxoWuzF4oiYHBF1EVHXq1evapdjZtZm5R0EKyTtB5D++0bOz29mZtvIOwimA+em0+cCj+T8/GZmto0sPz56P/A08GlJjZLOAyYBn5e0BPh8Om9mZlXUIasDR8TY7az6bFbPaWZmO6/FXiw2M7N8OAjMzArOQWBmVnAOAjOzgnMQmJkVnIPAzKzgHARmZgXnIDAzKzgHgZlZwTkIzMwKzkFgZlZwDgIzs4JzEJiZFZyDwMys4BwEZmYF5yAwMys4B4GZWcE5CMzMCs5BYGZWcA4CM7OCcxCYmRWcg8DMrOAcBGZmBecgMDMrOAeBmVnBdah2AZaz+jvzeZ66r+fzPGa2y9wjMDMrOAeBmVnBOQjMzArO1wgsG3ldizCzXeYegZlZwTkIzMwKzkFgZlZwDgIzs4JzEJiZFVxVgkDSiZJelPSSpCuqUYOZmSVyDwJJ7YFbgS8AA4GxkgbmXYeZmSWq0SMYBrwUEUsj4gPgAeCUKtRhZmZU5wtlBwCvlsw3Akduu5GkC4AL0tn3JL2YQ2070hN4s8o1tBRuiy3cFlu4LbbYhbYY35x19K1ko2oEgZpYFh9ZEDEZmJx9OZWRVB8RddWuoyVwW2zhttjCbbFFa2uLapwaagQOLJnvA7xWhTrMzIzqBMGzwEGS+knqBJwFTK9CHWZmRhVODUXEBkkXAb8G2gN3RMTCvOv4GFrMaaoWwG2xhdtiC7fFFq2qLRTxkdPzZmZWIP5msZlZwTkIzMwKzkGwjUqHv5D0ZUkhqdV8RGxnlWsLSeMkrZTUkD7+RzXqzEMlvxeSzpC0SNJCST/Pu8a8VPB7cVPJ78SfJK2uRp15qKAtPilptqT5khZIGlWNOsuKCD/SB8nF6z8D/YFOwHPAwCa26wo8CfwBqKt23dVqC2Ac8ONq19pC2uIgYD6wVzq/T7XrrlZbbLP9N0k+EFL12qv0ezEZmJBODwSWVbvuph7uEWyt0uEv/hn4V2BdnsXlzEOBbFFJW5wP3BoRbwNExBs515iXnf29GAvcn0tl+aukLQLYM53uRgv9zpSDYGtNDX9xQOkGkoYAB0bEjDwLq4KybZE6Le3yTpV0YBPr24JK2uJg4GBJT0n6g6QTc6suX5X+XiCpL9APmJVDXdVQSVt8H/iKpEbgVyQ9pBbHQbC1HQ5/IakdcBPwndwqqp5KhgL5JVATEYcCvwXuzryq6qikLTqQnB4aTvJX8O2SumdcVzVUNERM6ixgakR8mGE91VRJW4wF7oqIPsAo4Gfp+0iL0uIKqrJyw190BQYDj0taBhwFTG+jF4zLDgUSEasi4m/p7E+Aw3OqLW+VDIvSCDwSEesj4mXgRZJgaGt2ZoiYs2i7p4WgsrY4D3gQICKeBjqTDEjXojgItrbD4S8i4p2I6BkRNRFRQ3KxeHRE1Fen3EyVHQpE0n4ls6OBxTnWl6dKhkX5BTACQFJPklNFS3OtMh8VDREj6dPAXsDTOdeXp0ra4hXgswCSBpAEwcpcq6yAg6BERGwANg1/sRh4MCIWSrpO0ujqVpevCtvi4vSjks8BF5N8iqjNqbAtfg2skrQImA1cHhGrqlNxdnbi/8hY4IFIPy7TFlXYFt8Bzk//j9wPjGuJbeIhJszMCs49AjOzgnMQmJkVnIPAzKzgHARmZgXnIDAzKzgHgRWGpFPTEWP/vtq1mLUkDgIrkrHAHJIv/mRCUvusjm2WFQeBFYKkPYBjSb7yf1bJ8u9Kel7Sc5Impcs+Jem36bJ5kv5O0nBJM0r2+7Gkcen0MknXSJoDnC7pfEnPpvtPk/SJdLt9JT2cLn9O0jGS/lnSJSXHnSjp4lwaxSyV+83rzapkDPBfEfEnSW9JGgrsmy4/MiL+KmnvdNv7gEkR8bCkziR/MJUbWXVdRBwHIKlHRPwknf4BSfjcAvw78EREnJr2HPYgGZvmIeBH6WBkZ5EMb2yWGweBFcVY4OZ0+oF0vh1wZ0T8FSAi3pLUFTggIh5Ol60DkJoaaHIrU0qmB6cB0J3kzf7X6fKRwNfS434IvAO8I2lVOrz5vsD8tjg0hbVsDgJr8yT1IHkTHiwpSO4sFcA0Pjps8Pbe8Tew9anUztusX1syfRcwJiKeS08fDS9T4u0k4zT1Bu4os61Zs/M1AiuCLwP3RETfdOTYA4GXgbeA8SXn8PeOiDVAo6Qx6bLd0vXLgYHpfDfSESW3oyvwuqSOwDkly2cCE9Ljtpe06c5VDwMnAkewpfdglhsHgRXBWJI321LTgP1Jhg2ul9QAXJau+yrJyKoLgN8DvSPiVZJx5ReQXEOYv4Pnuxr4I/Ab4IWS5ZcAIyQ9D8wFBgGktzmcTTJ6ZVu9iYu1YB591KzK0ovE84DTI2JJteux4nGPwKyKJA0EXgJmOgSsWtwjMDMrOPcIzMwKzkFgZlZwDgIzs4JzEJiZFZyDwMys4P4/bga3Ylt58EYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "count    216.000000\n",
      "mean       0.694394\n",
      "std        0.098310\n",
      "min        0.436000\n",
      "25%        0.650250\n",
      "50%        0.725500\n",
      "75%        0.763000\n",
      "max        0.822000\n",
      "Name: test_accuracy, dtype: float64\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAHjVJREFUeJzt3XucVXW9//HXm5ugIiDgFWWg1DPcBETAS+cBWGqEhJk3SCVNTlppUZZWmnmCo+dnaccsH0ipeUlJNM30ZCFeIG8jjHAAlVSEUUNESUFJLp/fH3sxbHBg74FZe8/Mej8fj/1g7bW+a63PXjPs93zX2vu7FBGYmVl2tSh3AWZmVl4OAjOzjHMQmJllnIPAzCzjHARmZhnnIDAzyzgHgZWMpMsl3Zbi9hdIGpZMS9JNkt6V9IykT0l6Ma19mzVlDgJrUJLGSqqStFrSm5IeknR0KfYdEb0j4tHk6dHAZ4BuETE4Ip6IiEMaal+Sbpb0kwbYToWkkNSqIeoy2xEOAmswkiYC1wKTgb2BA4FfAp8vQzndgSURsWZnN9Tc36Sb++uzwhwE1iAkdQCuAL4WEfdExJqIWBcRf4yIi7axzu8l/UPSPyU9Lql33rKRkhZKel/S65K+k8zvIukBSaskvSPpCUktkmVLJH1a0jnAVOCIpGfyY0nDJNXkbX8/SdMlrZD0qqQL8pZdLuluSbdJeg8Yv1XdE4BxwHeT7f+xiG0OTnpK70laLulnyaLHk39XJds6oo7jtK11kXS0pL8lx2OZpPGbfh6SfpvU8pqkH+Ydp/GSZku6RtI7wOXJ/LMlLUpOp/1ZUvdkvpK2byU/q3mS+mzjV8Gaoojww4+dfgDHA+uBVttpczlwW97zs4H2wC7kehLVecveBD6VTHcCBibT/wXcALROHp8ClCxbAnw6mR4PzMrb3jCgJpluATwHXAa0AXoCrwDH5dW5DhiTtG1Xx2u5GfhJ3vNC23wSOCOZ3h0YmkxXAFHguG1r3QOB94HTk2PRGeifLPstcF9yfCuAl4Bz8o7NeuAbQCugXfJa/w5UJvN+CPwtaX9c8to6Akra7Fvu3zk/Gu7hHoE1lM7A2xGxvtgVIuI3EfF+RPyL3JvvoUnPAnJvxL0k7RER70bEnLz5+wLdI9fjeCKSd6t6OBzoGhFXRMRHEfEKcCNwWl6bJyPiDxGxMSI+bIBtrgM+KalLRKyOiKfqUe+21h0H/DUifpcci5URUS2pJXAqcElyfJcAPwXOyNvmGxFxXUSsT17ffwD/FRGLkp/hZKB/0itYRy5Q/o1c6C6KiDfrUb81cg4CaygrgS7Fnm+W1FLSlZJeTk6/LEkWdUn+PQkYCbwm6bG8Uyb/j9xfrg9LekXSxTtQa3dgv+R0yipJq4Dvk7uuscmyBt7mOcDBwAuSnpU0qh7b3ta6BwAv19G+C7leyWt5814D9s97vvXr6w78PK/2d8j99b9/RDwC/AK4HlguaYqkPepRvzVyDgJrKE8Ca8mdYijGWHIXkT8NdCB3+gJybz5ExLMR8XlgL+APwLRk/vsR8e2I6AmcAEyUdEw9a10GvBoRHfMe7SNiZF6bQr2MrZdvd5sRsTgiTk9ez1XA3ZJ2K2I/21t3GfCJOlZ5m9xf8d3z5h0IvF6g/v/Yqv52EfG3pIb/iYjDgN7kQqnO6z7WNDkIrEFExD/JnR+/XtIYSbtKai3ps5L+u45V2gP/IteT2JXcqQgAJLWRNE5Sh4hYB7wHbEiWjZL0SUnKm7+hnuU+A7wn6XuS2iW9kz6SDq/HNpaTuw5Q1DYlfUlS14jYCKxK1tkArAA2brWtLWxn3duBT0s6RVIrSZ0l9Y+IDeSCc5Kk9snpnYnA9r7DcQNwiZIL9snF5pOT6cMlDZHUGlhDLvDre8ytEXMQWIOJiJ+Re8P5Ibk3uGXA18n9Rb+135I7XfE6sBDY+pz5GcCS5LTRV4EvJfMPAv4KrCbXC/llbP7uQLF1biDXm+gPvEruL+ip5Homxfo1uWsYqyT9oYhtHg8skLQa+DlwWkSsjYgPgEnA7GRbQ+vY17bWXUru9Nm3yZ3KqQYOTdb5Brk37VeAWcAdwG+2c0zuJdfbuDM55v8HfDZZvAe56x3vkvuZrQSuLv5QWWO36dMWZmaWUe4RmJllnIPAzCzjHARmZhnnIDAzy7gmMdhUly5doqKiotxlmJk1Kc8999zbEdG1ULsmEQQVFRVUVVWVuwwzsyZF0muFW/nUkJlZ5jkIzMwyzkFgZpZxTeIaQV3WrVtHTU0Na9euLXcpVmJt27alW7dutG7dutylmDULTTYIampqaN++PRUVFeTGH7MsiAhWrlxJTU0NPXr0KHc5Zs1Ckz01tHbtWjp37uwQyBhJdO7c2T1BswbUZIMAcAhklH/uZg2rSQeBmZntvCZ7jWBrdzy9tEG3N3bIgQXbVFRU0L59e1q2bEmrVq126ktvI0eO5I477qBjx471Xre6upo33niDkSNHFm5cwOTJk/n+97+/09vZ5P7772fhwoVcfPGO3FHSzEqh2QRBucycOZMuXboUbljAgw8+uMPrVldXU1VVlWoQRAQRQYsW9etEjh49mtGjR+90XdYEVd1U7goa1qAvl7uC1PjUUMrGjx/Peeedx/Dhw+nZsyePPfYYZ599NpWVlYwfP762XUVFBW+//TZLliyhsrKSc889l969e3Psscfy4YcfAjBs2LDaXsfbb79NRUUFH330EZdddhl33XUX/fv356677mLNmjWcffbZHH744QwYMID77rsPgAULFjB48GD69+9Pv379WLx48Ra1XnzxxXz44Yf079+fcePG1dZy/vnnM3DgQJYtW8Z5553HoEGD6N27Nz/60Y+2qP9HP/oRAwcOpG/fvrzwwgsA3HzzzXz961+vPRYXXHABRx55JD179uTuu+8GYOPGjZx//vn07t2bUaNGMXLkyNplZpY+B8FOkMSxxx7LYYcdxpQpU7bZ7t133+WRRx7hmmuu4YQTTuBb3/oWCxYsYP78+VRXV3+s/eLFi/na177GggUL6NixI9OnT9/mttu0acMVV1zBqaeeSnV1NaeeeiqTJk1ixIgRPPvss8ycOZOLLrqINWvWcMMNN3DhhRfW9iC6deu2xbauvPJK2rVrR3V1NbfffjsAL774ImeeeSZz586le/fuTJo0iaqqKubNm8djjz3GvHnzatfv0qULc+bM4bzzzuPqq+u+k+Gbb77JrFmzeOCBB2pPF91zzz0sWbKE+fPnM3XqVJ588sltH3Qza3AOgp0we/Zs5syZw0MPPcT111/P448/Xme7E044AUn07duXvffem759+9KiRQt69+7NkiVLPta+R48e9O/fH4DDDjuszjbb8/DDD3PllVfSv39/hg0bxtq1a1m6dClHHHEEkydP5qqrruK1116jXbt2BbfVvXt3hg7dfBvdadOmMXDgQAYMGMCCBQtYuHBh7bIvfOELBWseM2YMLVq0oFevXixfvhyAWbNmcfLJJ9OiRQv22Wcfhg8fXq/Xa2Y7x0GwE/bbbz8A9tprL0488USeeeaZOtvtsssuALRo0aJ2etPz9evXb7M9QMuWLWvbtGrVio0bNwJs93P0EcH06dOprq6murqapUuXUllZydixY7n//vtp164dxx13HI888kjB17jbbrvVTr/66qtcffXVzJgxg3nz5vG5z31uizo21Z1f8/Ze26b7Zfu+2Wbl5SDYQWvWrOH999+vnX744Yfp06dPqvusqKjgueeeA9jiHHr79u1rawE47rjjuO6662rfYOfOnQvAK6+8Qs+ePbngggsYPXr0Fqd1NmndujXr1q2rc//vvfceu+22Gx06dGD58uU89NBDDfK6jj76aKZPn87GjRtZvnw5jz76aINs18yK02w+NVTMxz0b0vLlyznxxBMBWL9+PWPHjuX4449PdZ/f+c53OOWUU7j11lsZMWJE7fzhw4fXngq65JJLuPTSS/nmN79Jv379iAgqKip44IEHuOuuu7jtttto3bo1++yzD5dddtnH9jFhwgT69evHwIEDmTRp0hbLDj30UAYMGEDv3r3p2bMnRx11VIO8rpNOOokZM2bQp08fDj74YIYMGUKHDh0aZNtmVpiaQrd80KBBsfVn9BctWkRlZWWZKrKGtnr1anbffXdWrlzJ4MGDmT17Nvvss8822/vn3wT446NlJ+m5iBhUqF2z6RFY0zZq1ChWrVrFRx99xKWXXrrdEDCzhuUgsEbB1wXMyscXi83MMs5BYGaWcQ4CM7OMcxCYmWVc87lY3NAfVSvwUbFly5Zx5pln8o9//IMWLVowYcIELrzwwh3eXVMfhvorX/kKEydOpFevXttsc8MNN7Drrrty5pln7kyJZtbA3CPYQa1ateKnP/0pixYt4qmnnuL666/fYtyd+nrwwQd3KAQgFwQ7M4x1vsmTJ9c5PyJqh7eoy9SpU7cbAgBf/epXHQJmjZCDYAftu+++DBw4EMgN8VBZWcnrr7/+sXZZGYY6v7bdd9+dH/zgBxx66KEMHTq0dnC5yy+/vHZU0mHDhvG9732PwYMHc/DBB/PEE08A8MEHH3DKKafQr18/Tj31VIYMGbJTN/wxs8IcBA1gyZIlzJ07lyFDhtS5PCvDUG+yZs0ahg4dyvPPP8+///u/c+ONN9ZZ+/r163nmmWe49tpr+fGPfwzAL3/5Szp16sS8efO49NJLa8dWMrP0OAh20urVqznppJO49tpr2WOPPepsk5VhqDdp06YNo0aNKlh/XcNWz5o1i9NOOw2APn360K9fv/q8dDPbAc3nYnEZrFu3jpNOOolx48bVvqnVZWeHod50aqi+w1AfcsghW8yvrKxkyJAh/OlPf+K4445j6tSpWwxeV5e6hqF+9tln6dSpE+PHj6+zjtatWyOptv5CQ1Lnt2kKY1+ZNTfuEeygiOCcc86hsrKSiRMnlmSfzXUY6nxHH30006ZNA2DhwoXMnz+/wfdhZltqPj2CEo8MOHv2bG699Vb69u1bexpn8uTJDfIRzm1prsNQ5zv//PM566yz6NevHwMGDKBfv34ektosZR6G2hqVDRs2sG7dOtq2bcvLL7/MMcccw0svvUSbNm22aOeffxPgYajLzsNQW5P0wQcfMHz4cNatW0dE8Ktf/epjIWBmDctBYI1K+/bt/b0BsxJr0heLm8JpLWt4/rmbNawmGwRt27Zl5cqVflPImIhg5cqVtG3bttylmDUbqZ8aktQSqAJej4hRknoAdwJ7AnOAMyLio/put1u3btTU1LBixYqGLdgavbZt237sW9FmtuNKcY3gQmARsOlrt1cB10TEnZJuAM4BflXfjbZu3ZoePXo0XJVmZhmV6qkhSd2AzwFTk+cCRgCbvg11CzAmzRrMzGz70r5GcC3wXWDT+MWdgVURsWnMgRpg/5RrMDOz7UgtCCSNAt6KiPzhI1VH0zqv9kqaIKlKUpWvA5iZpSfNHsFRwGhJS8hdHB5BrofQUdKmaxPdgDfqWjkipkTEoIgY1LVr1xTLNDPLttSCICIuiYhuEVEBnAY8EhHjgJnAF5NmZwH3pVWDmZkVVo7vEXwPmCjp7+SuGfy6DDWYmVmiJENMRMSjwKPJ9CvA4FLstzG44+mlJdnP2CEHlmQ/Ztb8NNlvFpuZWcNwEJiZZZyDwMws4xwEZmYZ5yAwM8s4B4GZWcY5CMzMMs5BYGaWcQ4CM7OMcxCYmWWcg8DMLOMcBGZmGecgMDPLOAeBmVnGOQjMzDLOQWBmlnEOAjOzjCvJHcrMrBGpuqncFVgj4x6BmVnGOQjMzDLOQWBmlnEOAjOzjHMQmJllnIPAzCzjHARmZhnn7xE0E3c8vTT1fYwdcmDq+zCz0nOPwMws4xwEZmYZ5yAwM8s4B4GZWcY5CMzMMs5BYGaWcQ4CM7OMcxCYmWWcg8DMLONSCwJJbSU9I+l5SQsk/TiZ30PS05IWS7pLUpu0ajAzs8LS7BH8CxgREYcC/YHjJQ0FrgKuiYiDgHeBc1KswczMCkgtCCJndfK0dfIIYARwdzL/FmBMWjWYmVlhqV4jkNRSUjXwFvAX4GVgVUSsT5rUAPunWYOZmW1fqkEQERsioj/QDRgMVNbVrK51JU2QVCWpasWKFWmWaWaWaSX51FBErAIeBYYCHSVtGv66G/DGNtaZEhGDImJQ165dS1GmmVkmpfmpoa6SOibT7YBPA4uAmcAXk2ZnAfelVYOZmRWW5o1p9gVukdSSXOBMi4gHJC0E7pT0E2Au8OsUazAzswJSC4KImAcMqGP+K+SuF5iZNR1VN5VmP4O+XJr95PE3i83MMs5BYGaWcQ4CM7OMcxCYmWWcg8DMLOMcBGZmGVdUEEg6qph5ZmbW9BTbI7iuyHlmZtbEbPcLZZKOAI4EukqamLdoD6BlmoWZmVlpFPpmcRtg96Rd+7z577F5vCAzM2vCthsEEfEY8JikmyPitRLVZGZmJVTsWEO7SJoCVOSvExEj0ijKzMxKp9gg+D1wAzAV2JBeOWZmVmrFBsH6iPhVqpWYmVlZFPvx0T9KOl/SvpL23PRItTIzMyuJYnsEZyX/XpQ3L4CeDVuOmZmVWlFBEBE90i7EzMzKo6ggkHRmXfMj4rcNW46ZmZVasaeGDs+bbgscA8wBHARmZk1csaeGvpH/XFIH4NZUKjIzs5La0WGoPwAOashCzMysPIq9RvBHcp8Sgtxgc5XAtLSKssbpjqeXpr6PsUMOTH0fZralYq8RXJ03vR54LSJqUqjHzMxKrKhTQ8ngcy+QG4G0E/BRmkWZmVnpFHuHslOAZ4CTgVOApyV5GGozs2ag2FNDPwAOj4i3ACR1Bf4K3J1WYZZNvg5hVnrFfmqoxaYQSKysx7pmZtaIFdsj+F9JfwZ+lzw/FXgwnZLMzKyUCt2z+JPA3hFxkaQvAEcDAp4Ebi9BfWZmlrJCp3euBd4HiIh7ImJiRHyLXG/g2rSLMzOz9BU6NVQREfO2nhkRVZIqUqnILGWluCANvihtTUehHkHb7Sxr15CFmJlZeRQKgmclnbv1TEnnAM+lU5KZmZVSoVND3wTulTSOzW/8g4A2wIlpFmZmZqWx3SCIiOXAkZKGA32S2X+KiEdSr8zMGoWnX30n9X0M6eFboJdTsfcjmAnMTLkWMzMrg9S+HSzpAEkzJS2StEDShcn8PSX9RdLi5N9OadVgZmaFpTlMxHrg2xFRCQwFviapF3AxMCMiDgJmJM/NzKxMUguCiHgzIuYk0+8Di4D9gc8DtyTNbgHGpFWDmZkVVpKB45Ivnw0AniY3ZMWbkAsLYK9trDNBUpWkqhUrVpSiTDOzTEo9CCTtDkwHvhkR7xW7XkRMiYhBETGoa9eu6RVoZpZxqQaBpNbkQuD2iLgnmb1c0r7J8n2Bt7a1vpmZpS/NTw0J+DWwKCJ+lrfofuCsZPos4L60ajAzs8KKvR/BjjgKOAOYL6k6mfd94EpgWjJMxVJyt780M7MySS0IImIWuXsX1OWYtPZrZmb149tNmpllnIPAzCzj0rxGYNZsfGLp7+u/UssdGEht0Jfr1XxHbrLziaXpDyJXX6UY2A48uN22uEdgZpZxDgIzs4xzEJiZZZyDwMws43yxuJHYoYuRO+DlA/39PTPbknsEZmYZ5yAwM8s4B4GZWcb5GoFZY1J1U72aN8Yvh1nT4x6BmVnGOQjMzDLOQWBmlnEOAjOzjHMQmJllnIPAzCzjHARmZhnnIDAzyzgHgZlZxjkIzMwyzkFgZpZxDgIzs4xzEJiZZZyDwMws4xwEZmYZ5yAwM8s4B4GZWcb5DmVmKXn6Vd89zJoG9wjMzDLOQWBmlnEOAjOzjHMQmJllnIPAzCzjUgsCSb+R9Jak/8ubt6ekv0hanPzbKa39m5lZcdLsEdwMHL/VvIuBGRFxEDAjeW5mZmWUWhBExOPA1h+k/jxwSzJ9CzAmrf2bmVlxSn2NYO+IeBMg+XevbTWUNEFSlaSqFStWlKxAM7OsabQXiyNiSkQMiohBXbt2LXc5ZmbNVqmDYLmkfQGSf98q8f7NzGwrpQ6C+4GzkumzgPtKvH8zM9tKmh8f/R3wJHCIpBpJ5wBXAp+RtBj4TPLczMzKKLXRRyPi9G0sOiatfZqZWf012ovFZmZWGg4CM7OMcxCYmWWcg8DMLOMcBGZmGecgMDPLOAeBmVnGOQjMzDLOQWBmlnEOAjOzjHMQmJllXGpjDZmZNTZPv7r1TRMb3pAee6a+j4bmHoGZWcY5CMzMMs5BYGaWcQ4CM7OMcxCYmWWcg8DMLOMcBGZmGecgMDPLOAeBmVnGOQjMzDLOQWBmlnEOAjOzjHMQmJllnEcfzZhPLP19Sfbz8oEnl2Q/Zrbz3CMwM8s4B4GZWcY5CMzMMs7XCCwVpboWYWY7zz0CM7OMcxCYmWWcg8DMLOMcBGZmGecgMDPLuLIEgaTjJb0o6e+SLi5HDWZmllPyIJDUErge+CzQCzhdUq9S12FmZjnl6BEMBv4eEa9ExEfAncDny1CHmZlRni+U7Q8sy3teAwzZupGkCcCE5OlqSS+WoLbt6QK8XeYaGgsfi818LDbzsdhsJ47F2Q1ZR/diGpUjCFTHvPjYjIgpwJT0yymOpKqIGFTuOhoDH4vNfCw287HYrKkdi3KcGqoBDsh73g14owx1mJkZ5QmCZ4GDJPWQ1AY4Dbi/DHWYmRllODUUEeslfR34M9AS+E1ELCh1HTug0ZymagR8LDbzsdjMx2KzJnUsFPGx0/NmZpYh/maxmVnGOQjMzDLOQbCVYoe/kPRFSSGpyXxErL4KHQtJ4yWtkFSdPL5SjjpLoZjfC0mnSFooaYGkO0pdY6kU8XtxTd7vxEuSVpWjzlIo4lgcKGmmpLmS5kkaWY46C4oIP5IHuYvXLwM9gTbA80CvOtq1Bx4HngIGlbvuch0LYDzwi3LX2kiOxUHAXKBT8nyvctddrmOxVftvkPtASNlrL9PvxRTgvGS6F7Ck3HXX9XCPYEvFDn/xn8B/A2tLWVyJeSiQzYo5FucC10fEuwAR8VaJayyV+v5enA78riSVlV4xxyKAPZLpDjTS70w5CLZU1/AX++c3kDQAOCAiHihlYWVQ8FgkTkq6vHdLOqCO5c1BMcfiYOBgSbMlPSXp+JJVV1rF/l4gqTvQA3ikBHWVQzHH4nLgS5JqgAfJ9ZAaHQfBlrY7/IWkFsA1wLdLVlH5FDMUyB+BiojoB/wVuCX1qsqjmGPRitzpoWHk/gqeKqljynWVQ1FDxCROA+6OiA0p1lNOxRyL04GbI6IbMBK4NXkfaVQaXUFlVmj4i/ZAH+BRSUuAocD9zfSCccGhQCJiZUT8K3l6I3BYiWortWKGRakB7ouIdRHxKvAiuWBobuozRMxpNN/TQlDcsTgHmAYQEU8CbckNSNeoOAi2tN3hLyLinxHRJSIqIqKC3MXi0RFRVZ5yU1VwKBBJ++Y9HQ0sKmF9pVTMsCh/AIYDSOpC7lTRKyWtsjSKGiJG0iFAJ+DJEtdXSsUci6XAMQCSKskFwYqSVlkEB0GeiFgPbBr+YhEwLSIWSLpC0ujyVldaRR6LC5KPSj4PXEDuU0TNTpHH4s/ASkkLgZnARRGxsjwVp6ce/0dOB+6M5OMyzVGRx+LbwLnJ/5HfAeMb4zHxEBNmZhnnHoGZWcY5CMzMMs5BYGaWcQ4CM7OMcxCYmWWcg8AyQ9KJyYix/1buWswaEweBZcnpwCxyX/xJhaSWaW3bLC0OAssESbsDR5H7yv9pefO/K2m+pOclXZnM+6Skvybz5kj6hKRhkh7IW+8XksYn00skXSZpFnCypHMlPZusP13Srkm7vSXdm8x/XtKRkv5T0oV5250k6YKSHBSzRMlvXm9WJmOA/42IlyS9I2kgsHcyf0hEfCBpz6Tt7cCVEXGvpLbk/mAqNLLq2og4GkBS54i4MZn+CbnwuQ74H+CxiDgx6TnsTm5smnuAnyeDkZ1Gbnhjs5JxEFhWnA5cm0zfmTxvAdwUER8ARMQ7ktoD+0fEvcm8tQBSXQNNbuGuvOk+SQB0JPdm/+dk/gjgzGS7G4B/Av+UtDIZ3nxvYG5zHJrCGjcHgTV7kjqTexPuIynI3VkqgOl8fNjgbb3jr2fLU6ltt1q+Jm/6ZmBMRDyfnD4aVqDEqeTGadoH+E2BtmYNztcILAu+CPw2IronI8ceALwKvAOcnXcOf8+IeA+okTQmmbdLsvw1oFfyvAPJiJLb0B54U1JrYFze/BnAecl2W0radOeqe4HjgcPZ3HswKxkHgWXB6eTebPNNB/YjN2xwlaRq4DvJsjPIjaw6D/gbsE9ELCM3rvw8ctcQ5m5nf5cCTwN/AV7Im38hMFzSfOA5oDdAcpvDmeRGr2yuN3GxRsyjj5qVWXKReA5wckQsLnc9lj3uEZiVkaRewN+BGQ4BKxf3CMzMMs49AjOzjHMQmJllnIPAzCzjHARmZhnnIDAzy7j/DwPe2VNnRmiBAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#investigating the scores\n",
    "\n",
    "sns.distplot(df_2['test_accuracy'], kde=False, bins=10)\n",
    "plt.xlim([0.35, 0.85])\n",
    "plt.title('Classifier test scores')\n",
    "plt.xlabel('Accuracy')\n",
    "plt.ylabel('Count')\n",
    "plt.legend(['2 minutes training'], loc='upper right')\n",
    "plt.show()\n",
    "\n",
    "print(df_2['test_accuracy'].describe())\n",
    "\n",
    "sns.distplot(df_5['test_accuracy'], kde=False, bins=10, color=colors[1])\n",
    "plt.xlim([0.35, 0.85])\n",
    "plt.title('Classifier test scores')\n",
    "plt.xlabel('Accuracy')\n",
    "plt.ylabel('Count')\n",
    "plt.legend(['5 minutes traning'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "print(df_5['test_accuracy'].describe())\n",
    "\n",
    "sns.distplot(df_2['test_accuracy'], kde=False, bins=10)\n",
    "sns.distplot(df_5['test_accuracy'], kde=False, bins=10)\n",
    "plt.xlim([0.35, 0.85])\n",
    "plt.title('Classifier test scores')\n",
    "plt.xlabel('Accuracy')\n",
    "plt.ylabel('Count')\n",
    "plt.legend(['5 minutes traning', '2 minutes training'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "****************************************\n",
      "* Top 5% test scores 2 minutes training\n",
      "****************************************\n",
      "Total rows:10\n",
      "\n",
      "Scores\n",
      "[0.798 0.782 0.778 0.772 0.772 0.761 0.76  0.757 0.751 0.747]\n",
      "\n",
      "Parameter frequencies\n",
      "layers\n",
      "[16, 32]       2\n",
      "[16]           1\n",
      "[32]           1\n",
      "[8, 16, 32]    1\n",
      "[8, 16]        4\n",
      "[8]            1\n",
      "dtype: int64\n",
      "\n",
      "learning_rate\n",
      "0.05    1\n",
      "0.10    9\n",
      "dtype: int64\n",
      "\n",
      "num_hidden\n",
      "32     1\n",
      "64     2\n",
      "128    7\n",
      "dtype: int64\n",
      "\n",
      "use_dropout\n",
      "False    10\n",
      "dtype: int64\n",
      "\n",
      "use_max_pool\n",
      "True    10\n",
      "dtype: int64\n",
      "\n",
      "****************************************\n",
      "* Top 5% test scores 5 minutes training\n",
      "****************************************\n",
      "Total rows:10\n",
      "\n",
      "Scores\n",
      "[0.822 0.82  0.817 0.817 0.815 0.814 0.812 0.811 0.81  0.81 ]\n",
      "\n",
      "Parameter frequencies\n",
      "layers\n",
      "[16, 32]       6\n",
      "[8, 16, 32]    3\n",
      "[8, 16]        1\n",
      "dtype: int64\n",
      "\n",
      "learning_rate\n",
      "0.05    6\n",
      "0.10    4\n",
      "dtype: int64\n",
      "\n",
      "num_hidden\n",
      "32     3\n",
      "64     3\n",
      "128    4\n",
      "dtype: int64\n",
      "\n",
      "use_dropout\n",
      "False    10\n",
      "dtype: int64\n",
      "\n",
      "use_max_pool\n",
      "True    10\n",
      "dtype: int64\n",
      "\n"
     ]
    }
   ],
   "source": [
    "#investigating the top 5%\n",
    "\n",
    "def show_parameter_counts(df, percent, mins):\n",
    "    df_top = df.sort_values('test_accuracy', ascending=False).iloc[:int(df.shape[0] * percent)]\n",
    "    print('*' * 40)\n",
    "    print(\"* Top %d%% test scores %d minutes training\" % (percent * 100., mins))\n",
    "    print('*' * 40)\n",
    "    print(\"Total rows:%d\\n\" % df_top.shape[0])\n",
    "    print(\"Scores\")\n",
    "    print(df_top['test_accuracy'].values)\n",
    "    print('')\n",
    "    print('Parameter frequencies')\n",
    "    plot_columns = ['layers',\n",
    "                    'learning_rate',\n",
    "                    'num_hidden',\n",
    "                    'use_dropout',\n",
    "                    'use_max_pool']\n",
    "    for a in plot_columns:\n",
    "        print(df_top.groupby(a).size())\n",
    "        print('')\n",
    "        \n",
    "\n",
    "for a,b in [(df_2, 2), (df_5, 5)]:\n",
    "    show_parameter_counts(a, 0.05, b)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['learning_rate', 'step', 'loss', 'batch_accuracy',\n",
       "       'validation_accuracy', 'test_accuracy'], dtype=object)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# investigating the top classifier\n",
    "\n",
    "df_train = pd.read_csv('top_classifier_training.csv')\n",
    "df_train.columns.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_plot_details(title, xlabel, ylabel):\n",
    "    plt.title(title)\n",
    "    plt.xlabel(xlabel)\n",
    "    plt.ylabel(ylabel)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(df_train['step'], df_train['loss'])\n",
    "add_plot_details('Training loss function', 'Step', 'Loss')\n",
    "plt.plot(df_train['step'], df_train[['batch_accuracy', 'validation_accuracy', 'test_accuracy']])\n",
    "plt.legend(['Train batch accuracy', 'Validation Accuracy', 'Test accuracy'], loc='upper left')\n",
    "add_plot_details('Accuracy while training', 'Step', 'Accuracy')\n",
    "# plt.plot(df_train['step'], df_train['learning_rate'])\n",
    "# add_plot_details('Learning rate', 'Step', 'Learning rate')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "count    216.000000\n",
      "mean       0.694394\n",
      "std        0.098310\n",
      "min        0.436000\n",
      "25%        0.650250\n",
      "50%        0.725500\n",
      "75%        0.763000\n",
      "max        0.822000\n",
      "Name: test_accuracy, dtype: float64\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\navneet\\desktop\\udacity\\ml\\deeplearning\\digicam\\myvenv\\lib\\site-packages\\matplotlib\\axes\\_axes.py:6462: UserWarning: The 'normed' kwarg is deprecated, and has been replaced by the 'density' kwarg.\n",
      "  warnings.warn(\"The 'normed' kwarg is deprecated, and has been \"\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Investigating significance against benchmark\n",
    "\n",
    "print(df_5['test_accuracy'].describe())\n",
    "\n",
    "sns.distplot(df_5['test_accuracy'], kde=False, bins=100)\n",
    "plt.xlim([0.35, 0.85])\n",
    "plt.title('Classifier test scores')\n",
    "plt.xlabel('Accuracy')\n",
    "plt.ylabel('Count')\n",
    "plt.legend(['5 minutes traning', '2 minutes training'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\users\\navneet\\desktop\\udacity\\ml\\deeplearning\\digicam\\myvenv\\lib\\site-packages\\matplotlib\\axes\\_axes.py:6462: UserWarning: The 'normed' kwarg is deprecated, and has been replaced by the 'density' kwarg.\n",
      "  warnings.warn(\"The 'normed' kwarg is deprecated, and has been \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "count    194.000000\n",
      "mean       0.721031\n",
      "std        0.060520\n",
      "min        0.579000\n",
      "25%        0.681250\n",
      "50%        0.731000\n",
      "75%        0.765750\n",
      "max        0.822000\n",
      "Name: test_accuracy, dtype: float64\n",
      "Ttest_1sampResult(statistic=0.1374186487194639, pvalue=0.8921456924188548)\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import scipy.stats\n",
    "\n",
    "df_sig = df_5[df_5.test_accuracy > 0.55]\n",
    "sample = df_5['test_accuracy'].sample(20)\n",
    "\n",
    "sns.distplot(sample, kde=False, bins=100)\n",
    "\n",
    "print(df_sig['test_accuracy'].describe())\n",
    "\n",
    "print(scipy.stats.ttest_1samp(sample, df_sig['test_accuracy'].mean()))"
   ]
  },
  {
   "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": "3.6.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}