Gradient Descent - Linear Regression.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Untitled3.ipynb",
      "provenance": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/FazilShah/Machine-Learning/blob/master/Gradient%20Descent%20-%20Linear%20Regression.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "63kcIaUW12SN",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "def compute_cost():\n",
        "  x = data[:,0]\n",
        "  y = data[:,1]\n",
        "  total_error = 0\n",
        "  m = len(x)\n",
        "  for i in range(m):\n",
        "    total_error += ((theta0 + theta1*x[i]) - y[i])**2\n",
        "  return total_error/(2*m)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zeDc4die3K1K",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def step_gradient(theta0, theta1, data, alpha):\n",
        "  new_theta0 = theta0\n",
        "  new_theta1 = theta1\n",
        "  theta0_gradient = 0\n",
        "  theta1_gradient = 0\n",
        "  x = data[:,0]\n",
        "  y = data[:,1]\n",
        "  m = len(x)\n",
        "  for i in range(m):\n",
        "    theta0_gradient += ((new_theta0 + new_theta1*x[i]) - y[i])\n",
        "    theta1_gradient += (((new_theta0 + new_theta1*x[i])-y[i])*x[i])\n",
        "  new_theta0 = new_theta0 - ((alpha)*(theta0_gradient))/(m)\n",
        "  new_theta1 = new_theta1 - ((alpha)*(theta1_gradient))/(m)\n",
        "  return [new_theta0, new_theta1]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7TR87LBa7yxy",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def gradient_runner(theta0, theta1, data, alpha, iterations):\n",
        "  for i in range(iterations):\n",
        "    theta0, theta1 = step_gradient(theta0, theta1, data, alpha)\n",
        "  return [theta0, theta1]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "61whlDyk8wPO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "data = np.genfromtxt('/ex1data1.txt', delimiter = ',')\n",
        "a,b = gradient_runner(0,0,data,0.01,1500)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HDVSUbGb-Wrd",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 266
        },
        "outputId": "058b24bb-8a8c-43aa-c83f-d323878e3dc4"
      },
      "source": [
        "plt.scatter(data[:,0], data[:,1], marker='x', color='r')\n",
        "formula = eval('a + b*data[:,0]')\n",
        "plt.plot(data[:,0], formula)\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXIAAAD5CAYAAAA6JL6mAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0\ndHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3deZxcVZk38N/TISwidMI0JGEJzSaL\nvKQhPWlAIBF8AwSMMOO+gagREbo7QV8CSLojw6bQSQcdRwQEHET0BcfIEoiIbEJIAkkAAQnQKDFk\nETpEJIZ0PfPHqZu6deveW7eq7lr1+34+9amuU8s96a48deq5zzlHVBVERJRdTUl3gIiIasNATkSU\ncQzkREQZx0BORJRxDORERBnHQE5ElHHblHuAiOwF4BYAowAogOtUtV9EegF8FcC6/EMvUtV7/F6r\npaVFW1tba+owEVGjWbp06XpV3dXr/rKBHMAWAOer6lMishOApSKyMH/fHFW9OmhnWltbsWTJkqAP\nJyIiACLymt/9ZQO5qq4GsDr/80YReR7AHuF0j4iIalVRjlxEWgEcDmBRvulcEVkhIjeKyMiQ+0ZE\nRAEEDuQi8n4AdwDoVtW3AfwQwH4A2mBG7Nd4PG+aiCwRkSXr1q1zewgREdUgUCAXkeEwQfxWVb0T\nAFR1jaoOqWoOwI8BTHB7rqpep6rtqtq+666euXoiIqpS2UAuIgLgBgDPq2qfrX2M7WGnA3g2/O4R\nEWWYc1HCiBYpDFK18iEAXwDwjIgsy7ddBOAzItIGU5I4AOBrkfSQiCiLenuBwUFgzhxAxATx6dOB\nESPMfSEKUrXyKABxucu3ZpyIqGGpmiDe329uz5ljgnh/P9DVZe4Xt7BanSAjciIiqoSICd6ACd5W\nQO/qKozQwzxcnBtLtLe3KycEEVHDUAWabKcic7mqgriILFXVdq/7udYKEVEUrJy43fTpkZzwZCAn\nIgqbFcStnHguZ677+yMJ5syRExGFTcRUp9hz4lbOfMQI5siJiDLDWZ1SZbUKc+RERElxBu2QR+IW\nBnIiooxjICeiZMQ0fb0RMJATUfx6e4urN6wqj5CnrjcKBnIiipd9+roVzK1SvcFBjsyrwPJDIopX\nzNPXGwHLD4koGSFNX28ELD8kovSJcfp6I2AgJ6J4xTx9PWnXPfwy+n/7UqTHYI6ciOIVx/T1kGZU\nVn94Rd/CP+Ha363c2nb2pH2x3TbDIjkec+RElIyogm2MO/M4qSpm/+aPuOkPA1vbdm/eHvPPOwYt\n79+u6tctlyPniJyIkhHF9PWYd+axDOUU3/rlctz59KqtbQeO2gm/OPsoNO8wPPTjOTGQE1H9iLm0\ncfOWHM65dSl++/zarW3te4/ELV+egPdtG194ZWqFiOpPxKWN724ewpk/eRKLXn1za9vxB+2GH37+\niEjy4EytEFFj8SptDGFEvnHTe/jkj57A86vf3tp2WtvuuPoT47DNsOSKABnIiah+2EsbOzuBuXML\nt1XN7SqC+ZvvbMbU7z+K1996d2vb548ci+9MPRRNTclPYmIgJ6L6YZU2dnQU2ubMMUF80SJg9uyK\nKldWvD6Iqd9/rKjtnEn74VsnHghJ0SxUBnIiqi89PcBbbwHz5hWf/Fy0CDjyyECVK4+tXI/PXb+o\nqO2Ckw7C1yftF1Wva8JATkT1RaSQQqmwcuXuFavxjZ89VdR20OidsKD7uCh7XDNWrRBRfaqgcuXW\nRa/h4l89W9T24QN3xU++NCHKHgbGqhUiajwBK1eufeAlXLPwT0UP+1T7Xrjq44fF0cvQMJATUX1x\nLspln90JAHPmoGf+c7j58deKnvaND++Hb514UAIdrl3ZQC4iewG4BcAoAArgOlXtF5FdANwOoBXA\nAIBPqupb0XWViCgAn0W5znnfeNxz4T1FD5916iE465h9kuhpaMrmyEVkDIAxqvqUiOwEYCmA0wCc\nCeBNVb1SRGYCGKmqF/i9FnPkRBQbW3XKv/3nY3jqz4NFd8/51DicfvieSfSsYjXnyFV1NYDV+Z83\nisjzAPYA8DEAk/IPuxnA7wH4BnIiotiIoHXm3SXNPznzX/Hhg3ZLoEPRqShHLiKtAA4HsAjAqHyQ\nB4A3YFIvRESJUlXs40ifAMAdXz8K4/feJYEeRS9wIBeR9wO4A0C3qr5tn9WkqioirjkaEZkGYBoA\njB07trbeEhF5GMop9ruoNIDf+pUOfGj/lgR6FJ9AgVxEhsME8VtV9c588xoRGaOqq/N59LVuz1XV\n6wBcB5gceQh9JiLa6t3NQzh41oKS9p99tQNH71ffAdwSpGpFANwA4HlV7bPdNR/AGQCuzF//OpIe\nEhG5ePOdzTji0oUl7fd1H4cDR++UQI+SE2RE/iEAXwDwjIgsy7ddBBPAfyEiXwbwGoBPRtNFIqKC\nP//tHzjuew+WtD9+4fEY07xDAj1KXpCqlUcBeC1OcEK43SEicue2EiEArOidjJ23j347tTTjzE4i\nSrUHX1iLL920uKT9xf84KbJd6bOGgZyIUun2xX/GBXc8U9L+yuVTUrGZQ5owkBNRqvQt/BPmPfBS\nSfvAlack0JtsYCAnolSYcfsy3Pn0qpJ2BvDyGMiJKFGn/eAxLPvLYEk7A3hwDORElAi3dVAABvBq\nMJATUawyFcCd+3sG2O8zCQzkRBSLTAVwAOjtBQYHC2uaWxtWjBhh7ksRBnIiilTmAjhggvbgYNGu\nQkW7DqVsZN54gTwjX5WIss4tgI/eeXs8cVEGJoTbdhVCf38hoNt3HUqRsjsEhSnxHYIy9FWJKIu8\n1gJP0470FVEFmpoKt3O5RIJ4zTsE1Y2MfVUiypItQznsf/G9Je1fO25fXDjl4AR6FAJroGc3fXoq\nR+SNE8gz9lWJKAv+/s8tOLTnvpL2y04/FJ/r2DuBHoXECuLWQM8+8ANSFzMaJ5ADhWBu/TGA1P1B\niLLgjQ2bcOQVD5S033zWBEz8wK7BXiTN56tETMrVPtCzBoIjRqSnn3mNFcgz9FWJKI2e++sGnDLv\n0ZL2Bd3H4qDROwd/oSycr+rtLf5wsYJ5CmNF4wTyjH1VIkqTB19ciy/9pHQp2ScvOgG77bx9ZS+W\npfNVzn6kpV8OjRPIM/ZViSgNfvr4AC759XMl7c/NPhE7bldl+OD5qtA1VvkhkO68HFFKfOc3f8SN\nj71a0v7y5VMwLKy1wFNS2pcFLD90yshXJaIkfP76RXh05fqS9tBnYfJ8VagaL5ATUYkjLl2IN9/Z\nXNIeyTR6nq8KHQM5UQNLZB0Unq8KXePlyIkoHQtZ8XxVYMyRE9FWqQjgFp6vCg0DOVEDSFUAp9Ax\nkBMFkdE0AAN4Y2AgJyonC9PJHdwCePMOw7G8Z3ICvaGoMZAT+cnSdHK4B/AP7f8vuPUrRybQG4oL\nAzmRnwxMJx/KKfa7qHQzhzOPbkXv1A8m0COKW9nyQxG5EcCpANaq6qH5tl4AXwWwLv+wi1S19J3k\nwPJDyqwUTif/x+YtOGRW6Vrgs6d+EGcc3Rp/hygyYZQf3gTg+wBucbTPUdWra+gbUTakbDr52rc3\nYcLlpWuB33hmO44/aFTs/aHklQ3kqvqwiLRG3xWiiNRScZKi6eQvvrERJ859uKT9rvOOwaF7NMfS\nB0qnWnLk54rIFwEsAXC+qr4VUp+IwlNrxUkKppM/8tI6fOGGJ0va/zDzeOw+YofIj0/pV20g/yGA\nSwFo/voaAGe5PVBEpgGYBgBjx46t8nBEVQir4iShnWJ+/uSfMfPOZ0ran519It5f7VrgVJcCrbWS\nT63cZZ3sDHqfE092UuzsqRFLiipO3Fxx7/P40UOvlLSvvOxkbDOsyeUZVO8iWWtFRMao6ur8zdMB\nPFvN6xBFLkMbbp9102L87oW1Je2vXjEFksL+UnqUDeQichuASQBaROR1AD0AJolIG0xqZQDA1yLs\nI1H1UlZx4uaoKx7A6g2bSto5jZ6CClK18hmX5hsi6AtRuFJUceKG66BQWHjGhOpXCipO3DCAU9i4\nsURaZXS1vVRKye+SAZyqxY0lsiiDq+0lzi9YJ7yBAQM4RY2BPG0yttpeKqT0g48BnOLCQJ42GVht\nL1VS+MHnFsC326YJL/7HybH2gxoHc+RplcLV9lIrJZN+3AL4aW27Y+6nD4+tD1SfyuXIGcjTKCWB\nKVMS+uDL5RT7uqwF3v2RA9D9kQ9EfnxqDDzZmTUpr31OpQQm/Wx6bwgHXbKgpP2aT4zDv4/fM5Jj\nEnlhIE+blNY+p1bMH3yD/9iMtu8sLGn/2Vc6cPT+LaEdh6gSDORplNBqe5kU0wffyrUb8ZG+0rXA\n7+s+DgeO3imUYxBVizlyqg8RTfr5w8vr8dkfLyppX3TRCRi18/Y1vz5REMyRU2MIedLP/1/6Or75\ny+Ul7St6J2Pn7YfX9NpEYWMgrycpmYqeZX33v4h5v1tZ0v7SZSdjeKVrgfPvQTFhIE+zSgJBSmc3\nZsXZP12KBc+9UdJe9Vrg/HtQjBjIKxHHCMt6TSsQ9PWZ+mi/QJDC2Y1ZMfF7D+K1v/2jpL2mafT8\ne1DMGMiDimOEZQ/eViB46CFg6lRgwwbvQMBp/RWLdB0U/j0oZgzkQcQxwnIeo6/PBPFly8wF8A8E\nGdrSLEmxLWTFvwfFiIE8iDhGWF7HsPM7Vga2NEtS7CsR8u9BMeKW3EHZA60l7P+Ubsewmz7dBAgn\n5+zGXM5c9/d7P6dBtM682zWID1x5SvRBnH8PiglH5EHFMcJyO0ZbG7B0KTBjhve0c07rL5HoWuD8\ne1DMGMiDiGM9D+cxmpuB+fNNfnzGDJMzB7wDAaf1A0jRZg78e1CMGMiDiGOE5XaMnh4TxEeMMCWI\n5QJBwluaJcktgI9833A8PWtyAr3Ja+C/B8WLa61UIs468iiPUSe81gI/9oAW/PTLHQn0iCgaXGsl\nTHGMsDiKK+vdzUM4eFbpWuCfGL8nvveJcQn0iChZDOSUGWve3oSOyx8oaf/2KQfjK8fum0CPiNKB\ngZxS79lVG3DqtY+WtN9wRjtOOHhUAj0iShcGckqtBc++gbP/e2lJ+71dx+LgMTsn0COidGIgp9T5\nz9+vxHcXvFjSvvjij2DXnbbzfiJPFFODKhvIReRGAKcCWKuqh+bbdgFwO4BWAAMAPqmqb0XXTWoE\nnbc9jfnL/1rS/sKlJ2H74cP8n8xlY6mBBZmifxOAkxxtMwE8oKoHAHggf5uoKsd990G0zry7JIi/\nesUUDFx5Svkgbl9wzJoCb02uGhzklHiqe2VH5Kr6sIi0Opo/BmBS/uebAfwewAUh9osaQGizMLls\nLDW4QBOC8oH8LltqZVBVR+R/FgBvWbddnjsNwDQAGDt27PjXXnstnJ5TZkU2jV7VzIC15HIM4lQX\nIp8QpKoqIp6fBqp6HYDrADOzs9bjUXZFug4Kl42lBlZtIF8jImNUdbWIjAGwNsxOUX2JfCGrOBY1\nI0qxagP5fABnALgyf/3r0HpUDZadpVKsu/EEXdSM7xWqQ2Vz5CJyG8yJzRYAawD0APgfAL8AMBbA\nazDlh2+WO1gki2ax7Cx1EltKtlyQ5nuFMqrmHLmqfsbjrhOq7lVYuFt5qiS+FrjfgmN8r1Ady/4y\ntvb8qIVlZ7FKPIAHxfcKZVS5EXn2AznAsrMEqCr2ubB0LXAghQHcju8VyqBygTz7my97lZ1xNl8k\nNr03hNaZd5cE8WP2b4l2Q+Mw8L1CdSrbi2ax7Cw2XmuBnz1xP8w8+aAEelQhvleojmU7kHO38sg9\n8/oGfPT7pWuBz/1UG047fI8EelQlvleojtVPjpy1waH69bJV6Pr5spL2O885GkeMHVn7AZL6m/G9\nQhnUGHt2cp/L0My8YwV+vvgvJe1/mHk8dh+xQzgHSbKem+8VqkP1EcipZh+++vd4df07Je3PzT4R\nO247LLyAx3puotAxkDc4rxrwVy47GU3DmsIfLXPJWaLQZb/8kCqTPyfSOvNu1yA+sGkhBq46FU3n\nz4hugwZ7MLcwiBNVjSPyRtLbi9ZN/+p619b6b51irqMcLXPJWaJQ1feI3DmCzMLEj4j63Drzbtcg\nPnDVqRjYtLBwnKhHy8567lzOXNu3aSOiitTviDyLK91F0GfPdVCuOtX84BxtVzNarqSkj/XcRKGr\nz0CexcqIkPvsGcCvmFK81ohbEK9k9mM1Hz69vcX/HiuYp+1vQpQR9RnIrcCgWpzr7eyMJ2BUM+mk\n1mqO/DF8VyIsN9q2Rsv235P1e3QbLdfy4cN6bqLQ1GcgB4DZs73bo0yteI1Qm5uL++QW5KzAaV9m\nNUgQD3QSM8Bo26lcvpqlhESpUJ8nO1WBt94C5s0rbp83z7S7BagwTjLaR6jWiTsrWM6fb07sWY+b\nPr3wgWIdq4rV+So6iemWm+7qMu2zZwPd3YXfW3e3uRx1lLntVX7IUkKi5KlqbJfx48drLHI51c5O\nVRN6ii+dneZ+u54e1a6uQnsuZ2739FR37K6u4mO2tZlr6xjW/V1dqrNmmeuhoUJ7W1uh3f68rYfI\n6d4X3OV62XpMx3OK+ue8be9TZ2fp787td+b37/U6NhFVBcAS9Ymt9ZlaEQFGjjS5XvuovLPTtEe5\nBZhbemTpUmDGjNL0Q19foR0w6Ze2NmDZMmDiRHM/sDU//c4/t+CDPfe5Htb3JKazf2637SkSp7lz\n3V+LS8MSpYNflA/7EtuIXNV9VO41sgxzVOn1WkNDxW3O0b/XsXM5fXntRtfR92G994Xb/1zO/VuM\n32uF+W2GiFyhzIi8PgO5M33hdtvtOV6BNozjWukVj2Dtdux7VvzVNYBPv/3p2v6tXn13++Cz2vxe\nyy1dQ0ShKRfI6ze1UsmkEytFYDd9eiH1EXRCjttx+/qAhx4y6RK39IN1DJsLv3Udbttmz5KXv/z0\n/4PPdoyt7d/qxvr3z5sHdHSYC2Bud3aai99rsZSQKFHp31jCmaOuJGcd5Ln2PK+Vn3ZeV1pO5zxO\nTw+wYYN7SeKGDebYnZ3Y532T4fbnuOPrR2H83rvU/m/1Yy+btFgTe3p6GJyJElRuY4l0B/K4ptlb\nx3GefATCq4n2CrQ+NeCLtluGUbMvru24laj1w4CIIlEukKe3jlx9arLDXFJV1QTyOXNM1YdVKWJx\nC+LOYwfpi0v6wasG/E/vmqVkR21YV9u/s9J+xp0iqeb3SEQl0psjj2PWYG+vmQBjldcNDQHtjg89\n54JRIXxLKLuQFVD7vzPti4alvX9EGZLeETngPWswDKrAggWFWYw9PcDo0SYnPno0sGWLyZHbvxHU\n+C3BczOHK08xdeDOf6dX7bbfbastjm8z1Up7/4gypqYRuYgMANgIYAjAFr8cTlVUzRRxu+5ucz1y\nZG0jt95eYNMm87NzKv/HP25y5dYJz+Zm94kzft8SbPll34WsrMcGWTo26Cg27WugpL1/RFnjV5tY\n7gJgAEBL0MdXVEfurGt2Th33mzYe5LWtWutx44prp90m8rg936/mPD9JxnMavVdf/OrAw6yNT4u0\n948oJZDZOnL7NHugeNTc0eE9bTzoa1vLszpH43ZeJzr9as6bm9H6zwnA9qUvN7BpYWlqKGgdeKWj\n2KCj/KSkvX9EWeIX5ctdALwK4CkASwFMK/f4qmZ2Wos6BRm5VTLD0G9hLa+p6W4zNW3XvgtZlZtl\nGbTvQX4XYc32jEra+0eUMoh4RH6Mqq4Skd0ALBSRF1T1YfsDRGQagGkAMHbsWLfXKC/M/LFFFbj9\ndvfjTZgAHHlk6eJP9tGzrea89YK7XF+mqArF3i+3XHqQ0r+go9i0b6eW9v4RZY1flK/kAqAXwDf9\nHlPxiDyq/LEzRz40pNrRYW4fdphZQtZv8af863mOwL0WsbrkkuoXmKo2R+53O2lp7x9RSiCqEbmI\n7AigSVU35n+eDOA7YXy42A4SfORWSf7YObJuagIef9xUxFibLFiv6XiuqmKfC+9x7e7A7y81o/nu\n+03uva0NOPZY4MknTZ9aWoD16wuv7bdcrvO29W+uZBSb9jVQ0t4/ooyoeoq+iOwL4Ff5m9sA+Jmq\nXub3nKrWWgH8g5xzjRD7mty5nH9wCBI88za8+x7Gzb7f9b6By08Gjj4aWLTINLS3mzr0ZcvM7fPO\nAx55pHj9Fos9MFvH90sTOdc98ekzEdWHclP0qx6Rq+orAMZV+/yKeI3c1DaxxO0Dqbvbv7olwIhw\n2V8GcdoPHnN9+sCmhabGfPr9ZiS+aBGw226A88Pq2mvNtfUNYNiwwn19fd4LaQGlI/dy/4YKPpyI\nqD6kt/zQzu8EoVsZob1k0XpMhcHsRw+9jCvufcH1voFNC83o+PL8CU9rudfzzisEbTcuS9Zi/Pji\nHYSsYA9UPlmG096JGlL6A3m54CRiRt32QD53rrm2cuHlgrjtg+Kj1z6KZ1ZtKHnIcUPrcct3v1g8\nOnbWdpczfnxhWdy+vsJta4TuzH/bX7NcELd/O7EeX8uWdUSUHX5nQsO+hF61MjTkv81ZkCqIMrMw\nf/bEQPlt1Ly2SLMu48YVatbb2gqzRavd/i3I76uS5xFRqiGzMzsB/9mMzc2FmmortaFaqBABCvln\nL6pmGVmXWZgL/vk4Duq71Dx/gsfo2MrLW+u/WJwplqlTTSWMiOl3U5N5rjPNYp8hWs2GxtWM5Iko\n89IdyAH34NRny013dJgg/sQTJoife65pa27eOmV+azmhjddCVs9f8+/Y4Rtnl6Zy7KyTkoOD5va8\neeZk5+bNwPDhpXnyt982187X9ArWzc3VTZbx6iuDOVF98xuuh32peoq+1670znYrdfHee8WpjEsu\n2fpyvtPo/VIcXpspd3QUL+hlb7f30ZniKLf7fKWTZTjtnahuIdOplXIj176+4pH6uHHm5OHw4ea2\ntb44gNbNHa6HGLj8ZHPS0c4+ivWahNPcDEycaI5v1Y9bKZ/jjiuUPXqNpnt7i09AOqtrKp0sw2nv\nRA0r3Xt2AqW7+KgWZmDa660BE8iXLy96uuc6KJsWFleOtLWVlgGWWx9F1X0Cktfj45DksYkoEtnd\ns7Oc++4rBNyhoZIg3nrBXa5BfOCKKSaI9/ebsj97EG9qMsG7q6t0CVk7v9y5W+CMM5By2jtRw0l/\namVwsHhiz/TpxSc5rfTBxInA8uXeI3BrNx6g9OSpFcSBYBOIyqV8eHKRiGKU7kBebjMF6zGqaN1h\nMnDB5JKXGLjqVDPivuwkMwJ3G0nPmFFYPMt6zXL9Yj6aiFIi/TlywDsXjTI70p97rlmsavlyE8yn\nTi3k1Z05cet+l1JF334xH01EEYts0azYeOSiW7f/v64PH9h2EfCb35gbixebYN3ebnLhEycCO+9c\nWIHQGok/9FDh/kqCMfPRRJQC6R6R23PRnZ3A3Llo9VoL/IophUCayxVy6RZ7GiSXK4zE3e4nIkqR\nciPydAdyAOjthS5YgH0mXVJy187YghVXTHVf4a9caaBPuoaIKE2yX37Y04NfHn16UdOMR/4bA1ed\nihXv/s7UlPf3m+oW+/xM5/on3d2FkXpPj3fpIBFRxqR/RA7gnU3v4fxLfoqZ138brYOrSx9gnagU\nMZOHrHVXDjsMWLECGDUKWLOmcG3lyJ2lg0yvEFEKZf9kJ4Adtx+O//rul4Crv+z+gGXLzLR4wOTF\nR482P1upkzVrCtednSYNM3EiSweJqC5kIpC7Vq7YtbUVgrJI4SSmfW9Mi32qv8UezImIMib9OXJn\n5Yq1jZudVUoIlA/IVi68t7c0Lz59OrdEI6LMSX8gt8+inDvX/NzWVvyYtjaz6iDgPXJvazMbPli5\n8Pnzzc9WMLc+LKyTpn6c9/MkKRElKP2BHDCjZGukvWGDGYF3dpoqlK4uc3twsFDBMm6ceayVK29p\nMY955BETzJ98srBYVn+/yaUHPdnpHMlbHwIcyRNRQrIRyIFC/ts+OhcBnn7aBOrmZmDkSBPg33gD\n2GMPYNUqc/uQQwrB/NprzfrhVqWLXSUbHFczkiciioLfrhNhX6raIciNtdvN0FBhR562NtUtW8xG\nx/bbXjsJ2Xfzce48VO7Y3OCYiGKEMjsEZaKOHID7AlWASa9Ya6lY2tqAJUuA8883I/hZswobSFha\nWoD166tbPIuzQokoRtmf2Qm456WPOgrYay8TgBcvLn78Rz9qgnh/v5kg1N1dWoq4fn0h3WItnmVN\nFNqwwTtN4rehBBFREvyG62Ffat582UphWBsdW5eWFvfUiZUqmTBBdfTo0rTKt7/tnyZx2wCZGxwT\nUcxQJrWS/hG5NVmnq6tQYTJvnjmJec455jHr17s/9+qrzWj5ySfNCVBnpcvGjWYkbtfXV7yVm70a\nxWtDCefWcNVgSSMRVammmZ0ichKAfgDDAFyvqleG0qvSAxXqxC3WGil+hg831x0d5uLc2b65uTCR\nyDJ+fOkmzPb8fG9v8e0gW8OV09trql7smz27rehIROTGb7jud4EJ3i8D2BfAtgCWAzjE7zlVV63Y\nq1O8Kk780ixDQ6VpD3s1i5WCcatiiTpdwnQNEZWBCFMrEwCsVNVXVHUzgJ8D+FhtHysuVM3o2JrA\nY3FWqVisk5h2zlE3YFI09jRJU5MZidvFsRKiW+qIKzESUQVqCeR7APiL7fbr+bZw2fPSzkC7++4m\n7+1sX7/ePH5oqBAg3SpLrBmjVjrDGfDjqkZxW7SLQZyIAor8ZKeITBORJSKyZN26ddW9SG+vOQnp\nDLQf/7h7e1ubaW9qKn8y0p6TtkbC1glRrw+AsLGkkYhq4Zd38bsAOArAfbbbFwK40O85VefI/fLI\nVl7bL78cJM/c01P6nK4u0x4l5siJqAyUyZHXUrWyGMABIrIPgFUAPg3gs7V9rHiw0iudncVlf6rA\n8uWl5YBA8Qg8SIoiimqUILxKGgFudEFEgVQdyFV1i4icC+A+mAqWG1X1udB6FtSkSWYPzjACsPM5\ncQXRpD5EiKgu1JQjV9V7VPUDqrqfql4WVqdcDmTqrOfNK151cN480+6UxQCY1IcIEWVeNrZ6s6cb\n+vsLW7l1dRVmYhIRNaj0T9G3uM3utCpWOPuRiBpYdgJ5Lme2Z7MbP772TR24xgkRZVw2Arnf7E6r\nZrya9Aq3bSOiOpCNQO43u7A0OtgAAAX0SURBVHPq1OJNHoKyTqBy2zYiyrhsnOwEzCg5lyudxWlt\nAlHpiNzvBCpL/4goQ7IxIgcK6ZUwp9FzjRMiqgPZCeRRbOrANU6IqA5kJ7UChDsD0rlQlrVRhZVi\n4ciciDIiW4EcCG8GJNc4IaI6IRpjGqG9vV2XLFkS2/ECcZ4orebEKRFRhERkqaq2e92f/hx51BN2\nuMYJEWVcugM5J+wQEZWV3kDOCTtERIGk92QnJ+wQEQWS/pOdqsVT8HM5BnEiaijZPtnJCTtERGWl\nN5AnvbM9EVFGpDtHzgk7RERlZSNHzgk7RNTAsp0jBzhhh4iojPQHciIi8sVATkSUcQzkREQZx0BO\nRJRxsVatiMg6AK9V+fQWAOtD7E7U2N/oZa3P7G+0stZfIHif91bVXb3ujDWQ10JElviV36QN+xu9\nrPWZ/Y1W1voLhNdnplaIiDKOgZyIKOOyFMivS7oDFWJ/o5e1PrO/0cpaf4GQ+pyZHDkREbnL0oic\niIhcpC6Qi8iAiDwjIstEpGSFLTHmichKEVkhIkck0c98Xw7M99O6vC0i3Y7HTBKRDbbHzIq5jzeK\nyFoRedbWtouILBSRl/LXIz2ee0b+MS+JyBkJ9/l7IvJC/m/+KxEZ4fFc3/dPjP3tFZFVtr/7FI/n\nniQiL+bfzzMT7O/ttr4OiMgyj+cm8fvdS0QeFJE/ishzItKVb0/l+9inv9G9h1U1VRcAAwBafO6f\nAuBeAALgSACLku5zvl/DALwBU+9pb58E4K4E+3UcgCMAPGtr+y6AmfmfZwK4yuV5uwB4JX89Mv/z\nyAT7PBnANvmfr3Lrc5D3T4z97QXwzQDvmZcB7AtgWwDLARySRH8d918DYFaKfr9jAByR/3knAH8C\ncEha38c+/Y3sPZy6EXkAHwNwixpPABghImOS7hSAEwC8rKrVTniKhKo+DOBNR/PHANyc//lmAKe5\nPPVEAAtV9U1VfQvAQgAnRdZRG7c+q+r9qrolf/MJAHvG0ZcgPH7HQUwAsFJVX1HVzQB+DvO3iZRf\nf0VEAHwSwG1R9yMoVV2tqk/lf94I4HkAeyCl72Ov/kb5Hk5jIFcA94vIUhGZ5nL/HgD+Yrv9er4t\naZ+G95v/KBFZLiL3isgH4+yUh1Gqujr/8xsARrk8Jq2/ZwA4C+ZbmZty7584nZv/Gn2jx9f+NP6O\njwWwRlVf8rg/0d+viLQCOBzAImTgfezor12o7+E07hB0jKquEpHdACwUkRfyI4jUEpFtAUwFcKHL\n3U/BpFv+ns+T/g+AA+Lsnx9VVRHJTOmSiFwMYAuAWz0ekpb3zw8BXArzn/JSmHTFWQn0o1Kfgf9o\nPLHfr4i8H8AdALpV9W2x7U2Qxvexs7+29tDfw6kbkavqqvz1WgC/gvn6abcKwF6223vm25J0MoCn\nVHWN8w5VfVtV/57/+R4Aw0WkJe4OOqyx0lH567Uuj0nd71lEzgRwKoDPaT6Z6BTg/RMLVV2jqkOq\nmgPwY49+pOp3LCLbAPg3ALd7PSap36+IDIcJireq6p355tS+jz36G9l7OFWBXER2FJGdrJ9hTg48\n63jYfABfFONIABtsX6+S4jmKEZHR+bwjRGQCzO/8bzH2zc18ANbZ+zMA/NrlMfcBmCwiI/Npgcn5\ntkSIyEkA/h+Aqar6D4/HBHn/xMJx3uZ0j34sBnCAiOyT/1b3aZi/TVI+AuAFVX3d7c6kfr/5/z83\nAHheVftsd6XyfezV30jfw1Geva3ibO++MGfulwN4DsDF+fazAZyd/1kA/ADmbP8zANoT7vOOMIG5\n2dZm7++5+X/LcpgTHEfH3L/bAKwG8B5MfvDLAP4FwAMAXgLwWwC75B/bDuB623PPArAyf/lSwn1e\nCZPrXJa//Ff+sbsDuMfv/ZNQf3+af3+ugAk4Y5z9zd+eAlPV8HKS/c2332S9b22PTcPv9xiYFNUK\n299/Slrfxz79jew9zJmdREQZl6rUChERVY6BnIgo4xjIiYgyjoGciCjjGMiJiDKOgZyIKOMYyImI\nMo6BnIgo4/4XsB9ZKZrTzHgAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}