deep learning addition.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Load layers\n",
    "from keras.layers import Input, Dense\n",
    "import matplotlib.image as img\n",
    "from scipy.cluster.vq import kmeans, vq, whiten\n",
    "import seaborn as sns, pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "from PIL import Image\n",
    "from keras.models import Sequential\n",
    "import tensorflow as tf\n",
    "from keras.models import Model\n",
    "from keras.utils import plot_model\n",
    "from sklearn.model_selection import train_test_split\n",
    "import numpy as np\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAD6CAYAAABamQdMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjMsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+AADFEAAAfp0lEQVR4nO3df2wc533n8ff3JOXu4t6ZcUzJrmxFdi07vyXbhCtfDnYsXVrLyVIOzpQt6VS6Z0dYXeJLrhZaG6HgAnKAFqWSXOJUOursi0rQZKM0jsnCuWtAJRAOkJJQOcpxolBWlVCWrXKVtGSLBoml+Ht/7Mx6djnL/c1dDT8vYMHdmWdmvs+zM9+dffaZobk7IiKSLP+i2QGIiEj9KbmLiCSQkruISAIpuYuIJJCSu4hIAim5i4gkUMOSu5ndbWYTZnbKzB5r1HZERGQ2a8Q4dzNbBJwEPgScBb4HbHb3H9V9YyIiMkujztxvA065+2l3fx0YAjY2aFsiIlJgcYPWuxx4JfL6LPDbxQpfeeWVvnLlygaFIiKSTMeOHfuZu7fHzWtUcreYaXn9P2a2HdgOsGLFCsbGxhoUiohIMpnZZLF5jeqWOQtcG3l9DfBatIC797l7h7t3tLfHfvCIiEiVGpXcvwesMrPrzOwtwAPAcIO2JSIiBRrSLePuF83sE8D/ARYBz7j7DxuxLRERma1Rfe64+wvAC41av4iIFKcrVEVEEqhhZ+4L1cjICP39/c0OQ0RibNu2jVQq1eww5oWSe50NDQ1x8OBBALq6upocjYgAuWNyyZIlCya5q1tGRCSBdObeAFu2bAFgYGCgyZGICMDWrVubHcK805m7iEgCKbmLiCSQkruISAIpuYuIJJCSu4hIAim5i4gkkJK7iEgCKbmLiCSQkruISAIpuYuIJJCSu4hIAim5i4gkkJK7iEgCKbmLiCRQ1cndzK41s2+Z2Qkz+6GZfTKYfoWZfdPMXg7+vq1+4YqISDlqOXO/CDzq7u8C1gIfN7N3A48Bo+6+ChgNXouIyDyqOrm7+zl3/37w/J+AE8ByYCNwICh2ALi31iBFRKQydelzN7OVwM3Ad4Bl7n4Osh8AwNJ6bENERMpXc3I3s98A/gr4lLv/YwXLbTezMTMbO3/+fK1hiIhIRE3J3cyWkE3sA+7+tWDylJldHcy/GsjELevufe7e4e4d7e3ttYSRSLt27WLXrl3NDkNELlG1jJYx4GnghLt/NjJrGOgOnncDz1cfnoiIVKOWM/cPANuAdWY2HjzuAf4E+JCZvQx8KHgtCTczM0P2876+jh8/jpnlPXbs2FHzegvXWfhodXHtfanXSeprcbULuvv/BYrtMeurXa9k7d69u9khVOTw4cMNWe93v/vdWdPuueeemtfr7szMzNDW1pabNj09zeWXX17zuudDXHu7O5lMthd02bJluemXUr2kfqpO7iKQPYME2L9/f0PWf9VVV+HuDVl3YcK7FBJgqfZeunT24LRLoV5Sf7r9gIhIAim5t5hMJkMmk2FoaIjOzs686UNDQ3nTR0ZGGBkZwczo7OzkzJkznDlzJlc+nB9dz/79+3P91jt27ODkyZN525+rnzZuXm9vL729vYyMjMwqU4uwLp2dnezatYujR49y9OjROZep5wij6HuwENo7NDMzk4s5fOzatSu3X+7Zs2dWXHv27GHPnj25dYSvo2Wi7RRdT9i+hw4dypsftmNnZyczMzPs2LFDI8gq5e5Nf9x6662eFFu2bMk9qpFKpTyVSjng2bfnzenhNMCPHDmSmzc5OemAp9NpT6fT7u55ZcPy4TLT09O5soBPTEzk1jU1NZW3XFS4nbh5cdNqMTw87MPDw7PqkUqlfGpqKnaZnp4e7+npqWg7xepzqbd3seVKCWOcmpryqamp2LqGdQunFxO+V+H7NTU1ldu/BwcH3d19dHTUR0dHHfDx8fFZ+3/YluPj43kxVKrW47JVAWNeJK/qzF1EJImKZf35fOjMfTbmOFsrnF5s3lzlx8fHfXx83AHv7e2taTvF4q2H6elpHx8fz52VA97X11e39VdT12LzWqm951rnXHp6emadHRdbV29vrwM+OTnpk5OTuelhXcOz89Dg4OCc9Yx+64puc3p6uqI6xFmIZ+7mDRqJUImOjg4fGxtrdhh1sXXr1tzzgYGBqtcT9qFG359ov2rh+xY3b67y9dxOsfU0wv79+xkZGWF4eLgu66umrsXmtVJ7lxPLXMI+8oMHD7Jz587YdR0/fpw1a9bQ19cHwMc+9jGAXP97V1cXK1asyJXv7OzM/VZQTCVtWYl6HZetxsyOuXtH3DwNhZRLyqZNm9i+fXuzw0i08AMUsj/gRpN71OrVq0mn07n3Y9OmTQCcOnUKIC+xA3mJvRVOKpNOfe4iIgmkM3chnU43O4SyXX755ZdUvHFaLf4dO3awd+9eAIaGhti+fTuTk5PA7LPvQul0mn379gHwjW98g8suu4zu7u6S2zx58iQ33nhjjZHLXJTcF6joeOt6XM4/X2ZmZujq6mp2GBVrxfYOrxu48847c9M2b94MlE7qobBrJlw2lUoV/T2kr68v14XT39/Pzp07c1fPZjIZ+vv7efTRR6urjMyibpkWE17EUfg6Og3evAy98Hm4TKHwAqiwfH9/P/39/aRSKVKpVF7ZdDqdO2CjSanwIqLoDbzCdUQvUqnW0NAQhw4dyl3YAm9e1HT48GHWrVsXu1ylF7kUtlv09aXa3nH7SpyjR49y++23c/vtt/Oud71r1nrD9i686Cpu3d3d3bmz9cK6RW3cuDH3/Mknn6StrS13kdOyZcvo6uoqO34pTcldRCSB1C3TYqJ384t7HWpra8uNOIje2TBcpnA0Qnh2Fg5HC4evxY2EePzxxwF49dVXuemmm3Jfs8Ozzi1btgDknUHv3r2bkZERvvjFLwLwyCOPlFHbeJdddhnr1795Y9Genh7uu+++XAz1EHe5fmE7Fs5r9fYudguCUrcmWLly5az1hjcme+SRR+jp6eFnP/sZAL/85S9nLb927dpcvNEunkJLly7N9eXv37+fJ598MveN5fHHH2fFihWxsXZ2dtZt6OtConHuddZK42nrPVZY5rZQ2zvspnrsscdyP8y2mlY6LutprnHu6pYREUkgdcuISE2+8pWvAFySo5iSTGfuCVU44kCjEBprIbV3OCopejvfM2fOFB3FJM1R85m7mS0CxoBX3f0jZnYdMARcAXwf2Obur9e6HalMsR9m57MvuNp7jF+K/dWt0N7zJToGvq+vL3dPGWkt9Thz/yRwIvL6T4HPufsq4B+Ah+qwDRERqUBNZ+5mdg3wYeAzwB9Y9lRtHbAlKHIA+GOgNX9CT7BWOGNshRjmy0Kqa3imrjP21lbrmfvngT8E3ghevx2YdveLweuzwPIatyEiIhWqOrmb2UeAjLsfi06OKRp7SmNm281szMzGzp8/X20YIiISo5Yz9w8AnWb2U7I/oK4jeybfZmZhd881wGtxC7t7n7t3uHtHe3t7DWGIiEihqpO7uz/u7te4+0rgAeCQu28FvgXcFxTrBp6vOUoREalII8a5/xHZH1dPke2Df7oB2xARkTnU5QpVd/828O3g+WngtnqsV0REqqMrVEVEEkjJXUQkgZTcRUQSSMldRCSBlNxFRBJIyV1EJIGU3EVEEkjJXUQkgZTcRUQSSMldRCSBlNxFRBJIyV1EJIHqcuMwyffss88CcOHChSZHIiIABw8eBGDLli0lSiaHknudPfDAA0rqCfHjH/849/yd73xnEyORWnV1dQHZ43OhULeMiEgCWSv81/aOjg4fGxtrdhgiebZu3Zp7PjAw0MRIROKZ2TF374ibpzN3EZEEUnIXEUkgJXcRkQSqKbmbWZuZfdXMfmxmJ8zsdjO7wsy+aWYvB3/fVq9gRUSkPLWeuf934H+7+zuB1cAJ4DFg1N1XAaPBaxERmUdVJ3cz+7fAHcDTAO7+urtPAxuBA0GxA8C9tQYpIiKVqeXM/XrgPPC/zOz/mdn/NLPLgGXufg4g+Ls0bmEz225mY2Y2dv78+RrCEBGRQrUk98XALcBed78Z+Gcq6IJx9z5373D3jvb29hrCEBGRQrUk97PAWXf/TvD6q2ST/ZSZXQ0Q/M3UFqKIiFSq6uTu7n8HvGJmNwWT1gM/AoaB7mBaN/B8TRGKiEjFar1x2CPAgJm9BTgN/D7ZD4yvmNlDwBmgq8ZtiIhIhWpK7u4+DsTd12B9LesVEZHa6ApVEZEEUnIXEUkgJXcRkQRSchcRSSAldxGRBFJyFxFJICV3EZEEUnIXEUkgJXcRkQRSchcRSSAldxGRBFJyFxFJICV3EZEEUnIXEUkgJXcRkQRSchcRSSAldxGRBFJyFxFJoJqSu5n9NzP7oZm9ZGaDZvavzOw6M/uOmb1sZn8Z/H9VERGZR1UndzNbDvxXoMPd3wssAh4A/hT4nLuvAv4BeKgegYqISPlq7ZZZDPxrM1sMvBU4B6wDvhrMPwDcW+M2RESkQlUnd3d/FegFzpBN6jPAMWDa3S8Gxc4Cy2sNUkREKlNLt8zbgI3AdcBvApcBG2KKepHlt5vZmJmNnT9/vtowREQkRi3dMv8B+Im7n3f3C8DXgH8HtAXdNADXAK/FLezufe7e4e4d7e3tNYQhIiKFaknuZ4C1ZvZWMzNgPfAj4FvAfUGZbuD52kIUEZFK1dLn/h2yP5x+H/hBsK4+4I+APzCzU8DbgafrEKeIiFRgcekixbn7E8ATBZNPA7fVsl4REamNrlAVEUkgJXcRkQRSchcRSSAldxGRBFJyFxFJICV3EZEEqmkopMil7tVXXwXgwx/+MG1tbXnzTp48mXv+wQ9+MG/e9PQ0hw4dAuCKK65obJAiVdCZu4hIAunMXRa0n//85wAcP358znLnzp2bNS0869eZu7QiJXdZ0N7//vcDcMMNN3Dq1Kmyl7vhhht43/ve16iwRGqmbhkRkQRSchcBHnzwQZYsWVJW2SVLlvDggw82NiCRGim5iwCbN2/mwoULZZW9cOECmzdvbnBEIrVRchcBrr/+em6++WbMjOy/J5gtnHfzzTdz/fXXz3OEIpVRchcRSSAld5FAd3c3ixYtYtGiRbHzw3nd3d3zHJlI5ZTcRQL3338/b7zxBm+88Ubs/HDe/fffP8+RiVRO49xFAldddRV33nknAIcPH+bXv/51bt6iRYu44447cuVEWl3JM3cze8bMMmb2UmTaFWb2TTN7Ofj7tmC6mdkXzOyUmb1oZrc0MngREYlXTrfMl4G7C6Y9Boy6+ypgNHgNsAFYFTy2A3vrE6bI/Ni6dStbt26teJ5IqzF3L13IbCXw1+7+3uD1BPBBdz9nZlcD33b3m8zsfwTPBwvLzbX+jo4OHxsbq60mInUwPT0NwNKlS/PGvS9ZsoRMJgMw6+6RIs1iZsfcvSNuXrU/qC4LE3bwd2kwfTnwSqTc2WCaiIjMo3qPlom7+iP2q4GZbTezMTMbO3/+fJ3DEKlOW1sbbW1tbNiwgcWLF+ceGzZsyM0TuRRUm9yngu4Ygr+ZYPpZ4NpIuWuA1+JW4O597t7h7h3t7e1VhiHSGNu2bePixYu5x7Zt25odkkhFqh0KOQx0A38S/H0+Mv0TZjYE/DYwU6q/vZiLFy9mVzg8nDckTWQ+vP7663mvf/WrX3Hw4MEmRSML1aJFi+js7ARg8eLK0nXJ0mY2CHwQuNLMzgJPkE3qXzGzh4AzQFdQ/AXgHuAU8Avg9yuKRkRE6qKs0TKNFjda5utf/zoAH/3oR5sRkohIS3juuecAuPfee2fNm2u0TMteofqLX/wi97wVPoBEROabmeXlwkro3jIiIgmk5C4ikkBK7iIiCaTkLiKSQEruIiIJpOQuIpJASu4iIgmk5C4ikkBK7iIiCaTkLiKSQEruIiIJpOQuIpJASu4iIgmk5C4ikkBK7iIiCaTkLiKSQEruIiIJpOQuIpJAJZO7mT1jZhkzeyky7c/M7Mdm9qKZPWdmbZF5j5vZKTObMLPfbVTgIiJSXDln7l8G7i6Y9k3gve7+fuAk8DiAmb0beAB4T7DMn5vZorpFKyIiZSn5D7Ld/bCZrSyY9jeRl0eB+4LnG4Ehd/8V8BMzOwXcBhypS7QJlslkOHToEM8++ywAw8PDAOzatStXZvfu3U2JrVxhHQCeffbZXB2kucJ9qNj+c/ToUQ4cOADAvn37SKfTpNNpAFavXh07HyCdTrN69epGhy9VKpncy/Cfgb8Mni8nm+xDZ4NpTWFmc85PpVK5vxs3bmTp0qXzEVasJ554gn379jVs/TMzMwC0tbXh7g3ZRq11OHo0u+scOHAgt550Ok1XVxe33nor0Nj4W0Wp/TbU29vLjTfeCMAdd9zB5ZdfXtF2wg/i9evXMzk5CcDevXsZGhrKfSB86lOfip0P2Q+Ncj/Ay6lTrfWRAu5e8gGsBF6Kmf5p4DnAgtdfAv5TZP7TwH8sss7twBgwtmLFCi80MDDgAwMDng2xelNTUz41NeXArHVNTk765OSk9/T0OOATExM+MTFR0/ZqEcZYa53jDA8P+/DwcEPWHVVtHY4cOZJbbnBwMG/e+Pi4p1IpT6VSDY+/lUT3W8Cnp6fz5kfbJZVK+dTUVEXrT6fTnk6n52zTUvMrUU19Kq1T0gC5XFhk/pgXydtVj5Yxs27gI8DWYCOQPVO/NlLsGuC1uOXdvc/dO9y9o729vdowREQkhr2Zl+colO1z/2t3f2/w+m7gs8Cd7n4+Uu49wLNk+9l/ExgFVrn7r+daf0dHh4+NjeVNC/uet27dWpev4eHXwrh1ZTIZli1blutL3Lt3b83bq0b0q2s96hyamZlh27ZtAIyMjDS0W6PaOuzYsSPXFRO33PHjxwFYs2ZN4rtlokq1ZyaTAeDhhx8GoL+/H6CsLo1y3qu5jptqVFofyNZpoXbRmBkDAwMAbNmyJW7+MXfviF242Cm9v9l9MgicAy6QPTN/CDgFvAKMB499kfKfBv4WmAA2lFq/u3PrrbfO+rpRr26ZECW6CijoTpiamsp1ZaRSKZ+envbp6WlPp9Pe09OTt+zU1JT39vZ6b2+vA7mvlaOjo7O2Mz097YODgz44OJhXdmJiIjaGsGwqlYpdV7i+6LJ9fX155cJup7hHtA5hPcJ5c9WhcLvF6lCusMsF8PHx8aLl5lpvOW0RVz66TF9fn/f19c3qDojbH8JujXL2h9HR0di2LKXc9hwdHXUgF2MYR7gPRfefYvtCuY9QT0/PrLrXuz5hnQqVOuaK1T3smgyXmZycnLXu6Hqj+0NczJUcN9Wghm6ZckbLbI6Z/PQc5T8DfKbUekVEpHHqMVrmkheOJAm7ZSD7tXBkZCT3+sSJE7ky0REhmUyGhx9+OPeVyd3zRiGMj4/nDRfbtm0by5dnBxBNT0/nvm6GIxCiCmMoFHa1rF69Ou8r7o4dO9i1a1du6Nvu3bt58sknc/OjZaN1gOxXv3D+oUOHcnUItxPd7vLly5mengay3QBxdSjX7t27c3Vds2YNfX19AGzatCnvK3lh7FHbtm0r2hbhNgrLhyOm3D2vHUZGRvK6A+L2h3B/CfeHaJdC4f6wfv16gFn7Q72Eo4leeOEFIDsCLFqXqLB9mtEtU66wPpCtU/g+QXnHXPieh3UPR2KlUqncyJ93vOMdLF++PK8bds+ePXR1dQHw6KOP5nJDb2/vrBgrPW7mXbFT+vl8zGe3zPj4eO5rf9i9EHZbHDlyxI8cOTJrGYJf9gt/3Xf33Ff6YtuMfm0NvxLGjcqZnp6O/coaNy26XWBWF8KRI0dmdeMUW0/huorVIaxH9GttuXUoV9gu4QiN8DE4OFi0/cP4wzoUa4vC9gi/9heOyAj3gXC7hW0R3R/miqNQuFyjujHmKjvXe1tq3dW+l5XGWEnZSo65ubZVrK3iRunEdctUctxUixq6ZZqe2H2ek3vco6enJ7aft5wdMdpXXOwRKjWsrJLkHt1upfWvpg7hctHEW8l25mqXOGGSLUz0cf2v1QyTLPZehB8iQEUfkIVxlGrHUvtKudssp2zSknslx1ylyT26v5U6qajkuKlWLcldNw4TEUmiYll/Ph+tMFpmrmXmWq6S9Za7LmLOPAqXKye2cstXsq5SZcuJt9L3Itq1QszZezXrrOZ9rbbupcrW470Jv23EdQXUsu5qjpu5VFqfsE7VxlTO+xU1MTER+w2st7e36rrUAp25N9/Jkyc5efLkvG0v+gNTOAa8HhpVh7idL8rMmJmZyf2AFbV27VqeeuopnnrqKQA6Ozvz5qdSqVx7lNsWYflMJpP7IbRQ9Af2SpXaH4odkNU6duwYAHfddRd33XVX1etpFWF9gKL1acQxd+ONNzI8PMzw8DDj4+O5++zs3LmTPXv2FF1uPo/9cim51ygc1dHf309/f39ecspkMnk7RFj2+PHjNSfkaHLft29f3nbPnDnDjh07yl5XGBeQq0O4vrAOYT2iZev5oQLZAzp6UEetWLGCFStWAPl1D1+H08K2COMP26KwPcKRFqdPn+b06dO56dFlw1ET5err6yu6P4QfInMliGqE6/385z9PKpVi3bp1rFu3rq7bmE+F9QnrFFXJMVep6EnG6tWr2bt3L3v37mV8fJydO3fGxhHGMtdx0wxK7iIiSVTs6+F8PhrZ5x4d/QDxw5ziFN7kqNxyhY/oFXCTk5MOb15NF50XvSIP8HQ6PWvd0dinpqaKjs5Ip9OzhilGy0av7iu3DmGsYR3CekTnxdWhXNHlRkdH80YpFF5JWjiyKWyXudoibthm3E23wu0Uxl7O/lB4k7pS+0MphUNLK71xWGE80f1nfHw8b92FbRQOGS42373yK1SrqU/csVpOGxfWvXB/KpYTiPxuUXjsFva7V3LcVIsa+tybnti9gcl9roYvtd7CsnFjpUPhnSXD8fLhZelxb+zk5GTe3fjC52FyCpPLXDtOKNwpw+2GMcTd2TJ6sPb09Mw6cArvkBnGVqwOYT0KP4zCdgrrUK5ovSYmJnKXfYfrn6tu0faIxl9O+cLthO1fmHji9odi4vaHSg70Uvtt+Ojt7Z11bUY166n0EaokudejPqXaOLq/zhV7qXlxtxSI+0E1jKPc46YatSR3dcuIiCRQWXeFbLT5uCukiMilppa7QurMXUQkgZTcRUQSSMldRCSBlNxFRBJIyV1EJIGU3EVEEqhkcjezZ8wsY2YvxczbaWZuZlcGr83MvmBmp8zsRTO7pRFBi4jI3Mo5c/8ycHfhRDO7FvgQcCYyeQOwKnhsB/YWLiciIo1XMrm7+2Hg72NmfQ74Q7KX3IY2An8RXBl7FGgzs6vrEqmIiJStqj53M+sEXnX3wnu+Lgdeibw+G0wTEZF5tLjSBczsrcCngd+Jmx0zLfbeAWa2nWzXTe4+3SIiUh/VnLn/FnAdcNzMfgpcA3zfzK4ie6Z+baTsNcBrcStx9z5373D3jvb29irCEBGRYipO7u7+A3df6u4r3X0l2YR+i7v/HTAM/F4wamYtMOPu5+obsoiIlFLOUMhB4Ahwk5mdNbOH5ij+AnAaOAXsB/5LXaIUEZGKlOxzd/fNJeavjDx34OO1hyUiIrXQFaoiIgmk5C4ikkBK7iIiCaTkLiKSQEruIiIJpOQuIpJASu4iIgmk5C4ikkAV3zisGQ4ePNjsEERELik6cxcRSaCWPXO/4YYbcs83bdrUxEhERJonmgsr0bLJ/bbbbgMge7saERGphLplREQSSMldRCSBlNxFRBJIyV1EJIGU3EVEEkjJXUQkgawVhhqa2Xngn4GfNTuWFnIlao9CapPZ1Cb5Flp7vMPd2+NmtERyBzCzMXfvaHYcrULtMZvaZDa1ST61x5vULSMikkBK7iIiCdRKyb2v2QG0GLXHbGqT2dQm+dQegZbpcxcRkfpppTN3ERGpk6YndzO728wmzOyUmT3W7Hiaxcx+amY/MLNxMxsLpl1hZt80s5eDv29rdpyNZGbPmFnGzF6KTIttA8v6QrDfvGhmtzQv8sYo0h5/bGavBvvJuJndE5n3eNAeE2b2u82JurHM7Foz+5aZnTCzH5rZJ4PpC3Y/Kaapyd3MFgFfAjYA7wY2m9m7mxlTk93l7msiQ7keA0bdfRUwGrxOsi8DdxdMK9YGG4BVwWM7sHeeYpxPX2Z2ewB8LthP1rj7CwDBcfMA8J5gmT8Pjq+kuQg86u7vAtYCHw/qvpD3k1jNPnO/DTjl7qfd/XVgCNjY5JhayUbgQPD8AHBvE2NpOHc/DPx9weRibbAR+AvPOgq0mdnV8xPp/CjSHsVsBIbc/Vfu/hPgFNnjK1Hc/Zy7fz94/k/ACWA5C3g/KabZyX058Erk9dlg2kLkwN+Y2TEz2x5MW+bu5yC7UwNLmxZd8xRrg4W873wi6GJ4JtJVt+Daw8xWAjcD30H7ySzNTu4WM22hDt/5gLvfQvZr5MfN7I5mB9TiFuq+sxf4LWANcA7YE0xfUO1hZr8B/BXwKXf/x7mKxkxLbLtENTu5nwWujby+BnitSbE0lbu/FvzNAM+R/Uo9FX6FDP5mmhdh0xRrgwW577j7lLv/2t3fAPbzZtfLgmkPM1tCNrEPuPvXgsnaTwo0O7l/D1hlZteZ2VvI/iA03OSY5p2ZXWZm/yZ8DvwO8BLZtugOinUDzzcnwqYq1gbDwO8FoyHWAjPh1/IkK+gv/ijZ/QSy7fGAmf1LM7uO7A+I353v+BrNzAx4Gjjh7p+NzNJ+Usjdm/oA7gFOAn8LfLrZ8TSpDa4HjgePH4btALyd7C//Lwd/r2h2rA1uh0GyXQ0XyJ5xPVSsDch+3f5SsN/8AOhodvzz1B79QX1fJJu4ro6U/3TQHhPAhmbH36A2+fdku1VeBMaDxz0LeT8p9tAVqiIiCdTsbhkREWkAJXcRkQRSchcRSSAldxGRBFJyFxFJICV3EZEEUnIXEUkgJXcRkQT6/3nomFja1ysZAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Input layer (from previous exercise)\n",
    "input_tensor = Input(shape=(1,))\n",
    "\n",
    "# Create a dense layer\n",
    "output_layer = Dense(1)\n",
    "\n",
    "# Connect the dense layer to the input_tensor\n",
    "\n",
    "output_layer= Dense(1, activation='linear',name=\"Predicted-Score-Diff\")\n",
    "output_tensor= output_layer(input_tensor)\n",
    "\n",
    "model= Model(input_tensor, output_tensor)\n",
    "\n",
    "model.compile(optimizer='adam',loss='mae')\n",
    "\n",
    "plot_model(model, to_file='model.png')\n",
    "img=plt.imread('model.png')\n",
    "plt.imshow(img)\n",
    "plt.show()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "353\n",
      "Train on 8616 samples, validate on 2154 samples\n",
      "Epoch 1/1\n",
      "8616/8616 [==============================] - 0s 16us/step - loss: nan - val_loss: nan\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": [
    "from numpy import array\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "from keras.layers.embeddings import Embedding\n",
    "from keras.layers import Dense\n",
    "from keras.layers import Flatten\n",
    "\n",
    "df=pd.read_csv('predictions.csv')\n",
    "#print(df.head())\n",
    "\n",
    "n_teams=pd.unique(df['team']).shape[0]\n",
    "\n",
    "print(n_teams)\n",
    "\n",
    "#embed_layer= Embedding(input_dim=n_teams\n",
    "#                       ,input_length=1\n",
    "#                      ,output_dim=1\n",
    "#                       ,name='Team-strength'\n",
    "#                      )\n",
    "\n",
    "#embed_tensor=embed_layer(input_tensor)\n",
    "\n",
    "#flatten_tensor=Flatten()(embed_tensor)\n",
    "\n",
    "#model=Model(input_tensor,flatten_tensor)\n",
    "#model.compile(optimizer='adam',loss='mae')\n",
    "\n",
    "#print(df[['team','opponent','team_score','opp_score','rank','opp_rank']])\n",
    "\n",
    "df['score_diff']=df['team_score']-df['opp_score']\n",
    "df['seed_diff']=df['rank']-df['opp_rank']\n",
    "\n",
    "#map=['V','H']\n",
    "data=df['location']\n",
    "values=array(data)\n",
    "label_encoder=LabelEncoder()\n",
    "integer_encoded=label_encoder.fit_transform(values)\n",
    "#print(integer_encoded)\n",
    "#onehot_encoder = OneHotEncoder(sparse=False)\n",
    "#integer_encoded = integer_encoded.reshape(len(integer_encoded), 1)\n",
    "#onehot_encoded = onehot_encoder.fit_transform(integer_encoded)\n",
    "#print(onehot_encoded)\n",
    "df['location']=integer_encoded\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(df['seed_diff'],df['score_diff'].values,test_size=0.3,random_state=42)\n",
    "\n",
    "model.fit(df['seed_diff']\n",
    "         ,df['score_diff']\n",
    "          ,batch_size=64\n",
    "    \n",
    "          ,validation_split=.20\n",
    "          ,verbose=True\n",
    "         )\n",
    "\n",
    "#model.fit(X_train\n",
    "#         ,y_train\n",
    "#          ,batch_size=64\n",
    "#          ,validation_split=.20\n",
    "#          ,verbose=True\n",
    "#         )\n",
    "\n",
    "plt.scatter(df['seed_diff'],df['score_diff'])\n",
    "plt.show()\n",
    "\n",
    "in_data_1=df['seed_diff']\n",
    "pred=model.predict(in_data_1)\n",
    "#pred=model.predict(X_test.values)\n",
    "#print(pred)\n",
    "\n",
    "#plt.scatter(X_test,pred)\n",
    "plt.scatter(df['seed_diff'],pred)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "10770/10770 [==============================] - 0s 17us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 1.4465    \n",
      "Epoch 2/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5420\n",
      "Epoch 3/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5452\n",
      "Epoch 4/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5437\n",
      "Epoch 5/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5453\n",
      "Epoch 6/10\n",
      "10770/10770 [==============================] - 0s 13us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5440\n",
      "Epoch 7/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5443\n",
      "Epoch 8/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5473\n",
      "Epoch 9/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5438\n",
      "Epoch 10/10\n",
      "10770/10770 [==============================] - 0s 12us/step - loss: nan - dense_2_loss: nan - dense_3_loss: 0.5473\n",
      "[array([[-0.08009687],\n",
      "       [-1.7059706 ]], dtype=float32), array([[-0.01455685],\n",
      "       [-0.28739512]], dtype=float32)]\n",
      "location            1.000000e+00\n",
      "team_score          7.020244e+01\n",
      "opp_score           7.020244e+01\n",
      "score_diff          0.000000e+00\n",
      "total_score         1.404049e+02\n",
      "game_id             1.580121e+01\n",
      "opp_game_id         1.580121e+01\n",
      "year                2.019613e+03\n",
      "month               5.480409e+00\n",
      "day                 1.511049e+01\n",
      "D1                  2.000000e+00\n",
      "OT                           NaN\n",
      "conf_game           6.298979e-01\n",
      "reg_season          9.552461e-01\n",
      "weights             8.571954e-01\n",
      "yusag_coeff         1.955246e-01\n",
      "rank                1.744294e+02\n",
      "off_coeff           5.746619e-02\n",
      "def_coeff           1.380584e-01\n",
      "opp_yusag_coeff     1.955246e-01\n",
      "opp_rank            1.744294e+02\n",
      "opp_off_coeff       5.746619e-02\n",
      "opp_def_coeff       1.380584e-01\n",
      "pred_score_diff    -4.222352e-17\n",
      "pred_team_score     7.099227e+01\n",
      "pred_opp_score      7.099227e+01\n",
      "pred_total_score    1.419845e+02\n",
      "wins                5.000000e-01\n",
      "seed_diff           0.000000e+00\n",
      "dtype: float64\n"
     ]
    }
   ],
   "source": [
    "# two input and two output network\n",
    "\n",
    "from keras.layers import Input, Dense\n",
    "from keras.optimizers import Adam\n",
    "from keras.models import Model\n",
    "from scipy.special import expit as sigmoid\n",
    "\n",
    "\n",
    "# Input layer (from previous exercise)\n",
    "input_tensor = Input(shape=(2,))\n",
    "\n",
    "# Create a dense layer\n",
    "output_tensor_1 = Dense(1, activation='linear', use_bias=False)(input_tensor)\n",
    "output_tensor_2 = Dense(1, activation='sigmoid', use_bias=False)(input_tensor)\n",
    "\n",
    "model= Model(inputs=input_tensor, outputs=[output_tensor_1,output_tensor_2])\n",
    "\n",
    "model.compile(loss=['mean_absolute_error', 'binary_crossentropy'], optimizer=Adam(0.01))\n",
    "\n",
    "plot_model(model, to_file='model.png')\n",
    "img=plt.imread('model.png')\n",
    "plt.imshow(img)\n",
    "plt.show()\n",
    "model.fit(df[['seed_diff','location']],\n",
    "          [\n",
    "              df[['score_diff']],df[['wins']]\n",
    "          ],\n",
    "            epochs=10,\n",
    "          verbose=True,\n",
    "          batch_size=64\n",
    "         )\n",
    "\n",
    "# Print the model weights\n",
    "print(model.get_weights())\n",
    "\n",
    "# Print the training data means\n",
    "print(df.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": "3.7.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}