Shop_Floor.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "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"
    },
    "colab": {
      "name": "Shop_Floor.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "toc_visible": true
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "e3Zwtc8byC6a",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "from xgboost import XGBClassifier\n",
        "from sklearn.metrics import matthews_corrcoef, roc_auc_score\n",
        "from sklearn.model_selection import cross_val_score, StratifiedKFold\n",
        "import matplotlib.pyplot as plt\n",
        "import seaborn as sns\n",
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "L3uubgecyC6e",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "date_chunks = pd.read_csv(\"train_date.csv\", index_col=0, chunksize=100000, dtype=np.float32)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZJNeoHYVyC6h",
        "colab_type": "code",
        "colab": {},
        "outputId": "933b44c0-d987-4742-fd74-c0163cb13171"
      },
      "source": [
        "for chunk in date_chunks:\n",
        "    print(chunk.head())\n",
        "    break"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "         L0_S0_D1     L0_S0_D3     L0_S0_D5     L0_S0_D7     L0_S0_D9  \\\n",
            "Id                                                                      \n",
            "4.0     82.239998    82.239998    82.239998    82.239998    82.239998   \n",
            "6.0           NaN          NaN          NaN          NaN          NaN   \n",
            "7.0   1618.699951  1618.699951  1618.699951  1618.699951  1618.699951   \n",
            "9.0   1149.199951  1149.199951  1149.199951  1149.199951  1149.199951   \n",
            "11.0   602.640015   602.640015   602.640015   602.640015   602.640015   \n",
            "\n",
            "        L0_S0_D11    L0_S0_D13    L0_S0_D15    L0_S0_D17    L0_S0_D19  \\\n",
            "Id                                                                      \n",
            "4.0     82.239998    82.239998    82.239998    82.239998    82.239998   \n",
            "6.0           NaN          NaN          NaN          NaN          NaN   \n",
            "7.0   1618.699951  1618.699951  1618.699951  1618.699951  1618.699951   \n",
            "9.0   1149.199951  1149.199951  1149.199951  1149.199951  1149.199951   \n",
            "11.0   602.640015   602.640015   602.640015   602.640015   602.640015   \n",
            "\n",
            "          ...       L3_S50_D4246  L3_S50_D4248  L3_S50_D4250  L3_S50_D4252  \\\n",
            "Id        ...                                                                \n",
            "4.0       ...                NaN           NaN           NaN           NaN   \n",
            "6.0       ...                NaN           NaN           NaN           NaN   \n",
            "7.0       ...                NaN           NaN           NaN           NaN   \n",
            "9.0       ...                NaN           NaN           NaN           NaN   \n",
            "11.0      ...                NaN           NaN           NaN           NaN   \n",
            "\n",
            "      L3_S50_D4254  L3_S51_D4255  L3_S51_D4257  L3_S51_D4259  L3_S51_D4261  \\\n",
            "Id                                                                           \n",
            "4.0            NaN           NaN           NaN           NaN           NaN   \n",
            "6.0            NaN           NaN           NaN           NaN           NaN   \n",
            "7.0            NaN           NaN           NaN           NaN           NaN   \n",
            "9.0            NaN           NaN           NaN           NaN           NaN   \n",
            "11.0           NaN           NaN           NaN           NaN           NaN   \n",
            "\n",
            "      L3_S51_D4263  \n",
            "Id                  \n",
            "4.0            NaN  \n",
            "6.0            NaN  \n",
            "7.0            NaN  \n",
            "9.0            NaN  \n",
            "11.0           NaN  \n",
            "\n",
            "[5 rows x 1156 columns]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JGj5EP4xyC6k",
        "colab_type": "text"
      },
      "source": [
        "The data for this competition represents measurements of parts as they move through Bosch's production lines. Each part has a unique Id. The goal is to predict which parts will fail quality control (represented by a 'Response' = 1).\n",
        "\n",
        "The dataset contains an extremely large number of anonymized features. Features are named according to a convention that tells you the production line, the station on the line, and a feature number. E.g. L3_S36_F3939 is a feature measured on line 3, station 36, and is feature number 3939.\n",
        "\n",
        "On account of the large size of the dataset, we have separated the files by the type of feature they contain: numerical, categorical, and finally, a file with date features. The date features provide a timestamp for when each measurement was taken. Each date column ends in a number that corresponds to the previous feature number. E.g. the value of L0_S0_D1 is the time at which L0_S0_F0 was taken.\n",
        "\n",
        "In addition to being one of the largest datasets (in terms of number of features) ever hosted on Kaggle, the ground truth for this competition is highly imbalanced. Together, these two attributes are expected to make this a challenging problem."
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CJ0gfkPayC6l",
        "colab_type": "text"
      },
      "source": [
        "**Description**\n",
        "\n",
        "train_numeric.csv - the training set numeric features (this file contains the 'Response' variable)\n",
        "\n",
        "test_numeric.csv - the test set numeric features (you must predict the 'Response' for these Ids)\n",
        "\n",
        "train_categorical.csv - the training set categorical features\n",
        "\n",
        "test_categorical.csv - the test set categorical features\n",
        "\n",
        "train_date.csv - the training set date features\n",
        "\n",
        "test_date.csv - the test set date features\n",
        "\n",
        "sample_submission.csv - a sample submission file in the correct format"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dnCdT5mQyC6m",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "num_chunks = pd.read_csv(\"train_numeric.csv\", index_col=0, usecols=list(range(969)), chunksize=100000, dtype=np.float32)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1X25tMZCyC6o",
        "colab_type": "code",
        "colab": {},
        "outputId": "ff8c3426-a541-4564-e4fe-27e72f1dfd1b"
      },
      "source": [
        "for chunk in num_chunks:\n",
        "    print(print(chunk.head()))\n",
        "    break"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "      L0_S0_F0  L0_S0_F2  L0_S0_F4  L0_S0_F6  L0_S0_F8  L0_S0_F10  L0_S0_F12  \\\n",
            "Id                                                                             \n",
            "4.0      0.030    -0.034    -0.197    -0.179     0.118      0.116     -0.015   \n",
            "6.0        NaN       NaN       NaN       NaN       NaN        NaN        NaN   \n",
            "7.0      0.088     0.086     0.003    -0.052     0.161      0.025     -0.015   \n",
            "9.0     -0.036    -0.064     0.294     0.330     0.074      0.161      0.022   \n",
            "11.0    -0.055    -0.086     0.294     0.330     0.118      0.025      0.030   \n",
            "\n",
            "      L0_S0_F14  L0_S0_F16  L0_S0_F18      ...       L3_S50_F4243  \\\n",
            "Id                                         ...                      \n",
            "4.0      -0.032      0.020      0.083      ...                NaN   \n",
            "6.0         NaN        NaN        NaN      ...                NaN   \n",
            "7.0      -0.072     -0.225     -0.147      ...                NaN   \n",
            "9.0       0.128     -0.026     -0.046      ...                NaN   \n",
            "11.0      0.168     -0.169     -0.099      ...                NaN   \n",
            "\n",
            "      L3_S50_F4245  L3_S50_F4247  L3_S50_F4249  L3_S50_F4251  L3_S50_F4253  \\\n",
            "Id                                                                           \n",
            "4.0            NaN           NaN           NaN           NaN           NaN   \n",
            "6.0            NaN           NaN           NaN           NaN           NaN   \n",
            "7.0            NaN           NaN           NaN           NaN           NaN   \n",
            "9.0            NaN           NaN           NaN           NaN           NaN   \n",
            "11.0           NaN           NaN           NaN           NaN           NaN   \n",
            "\n",
            "      L3_S51_F4256  L3_S51_F4258  L3_S51_F4260  L3_S51_F4262  \n",
            "Id                                                            \n",
            "4.0            NaN           NaN           NaN           NaN  \n",
            "6.0            NaN           NaN           NaN           NaN  \n",
            "7.0            NaN           NaN           NaN           NaN  \n",
            "9.0            NaN           NaN           NaN           NaN  \n",
            "11.0           NaN           NaN           NaN           NaN  \n",
            "\n",
            "[5 rows x 968 columns]\n",
            "None\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DKnTdzUmyC6s",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X = pd.concat([pd.concat([dchunk, nchunk], axis=1).sample(frac=0.05)\n",
        "               for dchunk, nchunk in zip(date_chunks, num_chunks)])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1DnGGEfUyC6u",
        "colab_type": "code",
        "colab": {},
        "outputId": "5211a979-351f-4197-c3d4-b59cfddc5500"
      },
      "source": [
        "y = pd.read_csv(\"train_numeric.csv\", index_col=0, usecols=[0,969], dtype=np.float32).loc[X.index].values.ravel()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\numpy\\lib\\arraysetops.py:472: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
            "  mask |= (ar1 == a)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9-iDtD7wyC6w",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "X = X.values"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "I2WEXvNIyC6z",
        "colab_type": "code",
        "colab": {},
        "outputId": "90e005bb-c7b7-4c02-a7d9-0647cf2f7e0b"
      },
      "source": [
        "y"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([0., 0., 0., ..., 0., 0., 0.], dtype=float32)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 9
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TmSF_QdLyC62",
        "colab_type": "code",
        "colab": {},
        "outputId": "f4d4341a-0f21-4dd3-fccb-acd825f5656a"
      },
      "source": [
        "y.sum()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "317.0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "21D-itHJyC64",
        "colab_type": "code",
        "colab": {},
        "outputId": "56c4b221-2df7-4d4d-e451-a2891e78d114"
      },
      "source": [
        "len(y)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "54187"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VvCOEBk-yC68",
        "colab_type": "code",
        "colab": {},
        "outputId": "6b5c1d51-9f8e-42a0-8f8f-530660f669b4"
      },
      "source": [
        "y.sum()/len(y)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.005850111650395852"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 12
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "X6VWo9dNyC6-",
        "colab_type": "code",
        "colab": {},
        "outputId": "7121e6eb-f3ec-4819-8a53-db4dcab8bccb"
      },
      "source": [
        "clf = XGBClassifier(base_score=0.005)\n",
        "clf.fit(X, y)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "XGBClassifier(base_score=0.005, booster='gbtree', colsample_bylevel=1,\n",
              "       colsample_bytree=1, gamma=0, learning_rate=0.1, max_delta_step=0,\n",
              "       max_depth=3, min_child_weight=1, missing=None, n_estimators=100,\n",
              "       n_jobs=1, nthread=None, objective='binary:logistic', random_state=0,\n",
              "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
              "       silent=True, subsample=1)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ibepTehcyC7A",
        "colab_type": "text"
      },
      "source": [
        "### Save Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_vYRQrJTyC7C",
        "colab_type": "code",
        "colab": {},
        "outputId": "845802b1-c637-414c-da8d-3751513a7f4c"
      },
      "source": [
        "import sys\n",
        "print(sys.version)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "3.6.4 |Anaconda, Inc.| (default, Jan 16 2018, 10:22:32) [MSC v.1900 64 bit (AMD64)]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hy6cZXn0yC7G",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import pickle"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4-YCxP-QyC7J",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# save the model to disk\n",
        "filename = 'XGBoost_Model.sav'\n",
        "pickle.dump(clf, open(filename, 'wb'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tsMbRcS6yC7P",
        "colab_type": "text"
      },
      "source": [
        "### Load Model"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "58ju7I-8yC7R",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# load the model from disk\n",
        "myModel = pickle.load(open(filename, 'rb'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MSm0Si-PyC7S",
        "colab_type": "code",
        "colab": {},
        "outputId": "dd279a75-af6a-4ad8-f2b9-51c4921d95c7"
      },
      "source": [
        "# threshold for a manageable number of features\n",
        "plt.hist(myModel.feature_importances_[myModel.feature_importances_>0])\n",
        "important_indices = np.where(myModel.feature_importances_>0.005)[0]\n",
        "print(important_indices)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[  50   76  393  883  939 1018 1019 1029 1156 1169 1171 1173 1178 1194\n",
            " 1196 1202 1203 1210 1219 1227 1228 1233 1236 1245 1249 1254 1263 1277\n",
            " 1297 1310 1408 1516 1542 1549 1550 1618 1837 1838 1843 1846 1879 1880\n",
            " 1885 1889 1890 1892 1893 1899 1911 1927 1932 1944 1959 1966 1968 1974\n",
            " 1983 1984 1985 1987 1988 1990 1994 1995 1997 2006 2007 2010 2018 2022\n",
            " 2028 2084 2093 2107]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAD8CAYAAAB5Pm/hAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAADzVJREFUeJzt3X2sJXddx/H3xy5tpYqF7i2W3YbdJht0QQ14rUWUEFahpYRdI5glBDbYZCXWR0ykFU0TE5I2GnkIitlQYEmQtlZINwJKU2gIRrbepYU+UbqUtb20spfwIA8RXPn6x5mLh+XuPXfPnHMffrxfycmZ+c1vZr6/nbufnTvnzGyqCklSu35krQuQJE2XQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklq3Ka1LgBg8+bNtW3btrUuQ5I2lCNHjnypqmZG9VsXQb9t2zbm5ubWugxJ2lCS/MdK+nnpRpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGrcu7oztY9tVH1izfR+79vI127ckrZRn9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaNzLok7wjyfEk9wy1/WWSzyT5dJL3Jzl3aNnVSY4meSDJC6dVuCRpZVZyRv8u4NKT2m4FnlFVPwt8FrgaIMlOYC/w9G6dv01yxsSqlSSdtpFBX1UfA758UtuHq+pEN/sJYGs3vRu4oaq+XVWfB44CF0+wXknSaZrENfrfAj7UTW8BHhlaNt+1/YAk+5PMJZlbWFiYQBmSpKX0CvokrwdOAO9ZbFqiWy21blUdqKrZqpqdmZnpU4YkaRljP48+yT7gxcCuqloM83ngwqFuW4FHxy9PktTXWGf0SS4FXge8pKq+NbToELA3yVlJtgM7gDv6lylJGtfIM/ok7wWeB2xOMg9cw+BbNmcBtyYB+ERVvaaq7k1yE3Afg0s6V1bV/06reEnSaCODvqpevkTz9cv0fwPwhj5FSZImxztjJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWrcyKBP8o4kx5PcM9T2pCS3Jnmwe39i154kb0lyNMmnkzxrmsVLkkZbyRn9u4BLT2q7CritqnYAt3XzAJcBO7rXfuBtkylTkjSukUFfVR8DvnxS827gYDd9ENgz1P7uGvgEcG6SCyZVrCTp9I17jf7JVfUYQPd+fte+BXhkqN981/YDkuxPMpdkbmFhYcwyJEmjTPrD2CzRVkt1rKoDVTVbVbMzMzMTLkOStGjcoP/i4iWZ7v141z4PXDjUbyvw6PjlSZL6GjfoDwH7uul9wC1D7a/qvn1zCfC1xUs8kqS1sWlUhyTvBZ4HbE4yD1wDXAvclOQK4GHgZV33DwIvAo4C3wJePYWaJUmnYWTQV9XLT7Fo1xJ9C7iyb1GSpMnxzlhJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGtcr6JP8UZJ7k9yT5L1Jzk6yPcnhJA8muTHJmZMqVpJ0+sYO+iRbgN8HZqvqGcAZwF7gOuCNVbUD+ApwxSQKlSSNp++lm03AjybZBDweeAx4PnBzt/wgsKfnPiRJPYwd9FX1BeCvgIcZBPzXgCPAV6vqRNdtHtiy1PpJ9ieZSzK3sLAwbhmSpBH6XLp5IrAb2A48BTgHuGyJrrXU+lV1oKpmq2p2ZmZm3DIkSSP0uXTzq8Dnq2qhqv4HeB/wS8C53aUcgK3Aoz1rlCT10CfoHwYuSfL4JAF2AfcBHwVe2vXZB9zSr0RJUh99rtEfZvCh6yeBu7ttHQBeB7w2yVHgPOD6CdQpSRrTptFdTq2qrgGuOan5IeDiPtuVJE2Od8ZKUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1LheQZ/k3CQ3J/lMkvuTPDvJk5LcmuTB7v2JkypWknT6+p7Rvxn456r6KeDngPuBq4DbqmoHcFs3L0laI2MHfZInAM8Frgeoqu9U1VeB3cDBrttBYE/fIiVJ4+tzRn8RsAC8M8mdSd6e5BzgyVX1GED3fv5SKyfZn2QuydzCwkKPMiRJy+kT9JuAZwFvq6pnAt/kNC7TVNWBqpqtqtmZmZkeZUiSltMn6OeB+ao63M3fzCD4v5jkAoDu/Xi/EiVJfYwd9FX1n8AjSZ7WNe0C7gMOAfu6tn3ALb0qlCT1sqnn+r8HvCfJmcBDwKsZ/ONxU5IrgIeBl/XchySph15BX1V3AbNLLNrVZ7uSpMnxzlhJapxBL0mNM+glqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGtc76JOckeTOJP/UzW9PcjjJg0luTHJm/zIlSeOaxBn9HwD3D81fB7yxqnYAXwGumMA+JElj6hX0SbYClwNv7+YDPB+4uetyENjTZx+SpH76ntG/CfgT4Lvd/HnAV6vqRDc/D2zpuQ9JUg9jB32SFwPHq+rIcPMSXesU6+9PMpdkbmFhYdwyJEkj9Dmjfw7wkiTHgBsYXLJ5E3Bukk1dn63Ao0utXFUHqmq2qmZnZmZ6lCFJWs7YQV9VV1fV1qraBuwFPlJVrwA+Cry067YPuKV3lZKksU3je/SvA16b5CiDa/bXT2EfkqQV2jS6y2hVdTtwezf9EHDxJLYrSerPO2MlqXEGvSQ1zqCXpMYZ9JLUOINekhpn0EtS4wx6SWrcRL5H/8Nq21UfWJP9Hrv28jXZr6SNyTN6SWqcQS9JjTPoJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGjf28+iTXAi8G/hJ4LvAgap6c5InATcC24BjwG9W1Vf6l6pFa/UcfPBZ+NJG1OeM/gTwx1X108AlwJVJdgJXAbdV1Q7gtm5ekrRGxg76qnqsqj7ZTX8duB/YAuwGDnbdDgJ7+hYpSRrfRK7RJ9kGPBM4DDy5qh6DwT8GwPmT2IckaTy9gz7JjwH/CPxhVf3Xaay3P8lckrmFhYW+ZUiSTqFX0Cd5HIOQf09Vva9r/mKSC7rlFwDHl1q3qg5U1WxVzc7MzPQpQ5K0jLGDPkmA64H7q+qvhxYdAvZ10/uAW8YvT5LU19hfrwSeA7wSuDvJXV3bnwLXAjcluQJ4GHhZvxIlSX2MHfRV9XEgp1i8a9ztSpImyztjJalxBr0kNc6gl6TGGfSS1DiDXpIaZ9BLUuMMeklqnEEvSY0z6CWpcQa9JDXOoJekxvV5qJm0avx/cqXxeUYvSY0z6CWpcQa9JDXOoJekxhn0ktQ4g16SGmfQS1LjDHpJapw3TOm0rOWNS5LG4xm9JDXOM3pphLX6LcZHL2hSpnZGn+TSJA8kOZrkqmntR5K0vKkEfZIzgL8BLgN2Ai9PsnMa+5IkLW9al24uBo5W1UMASW4AdgP3TWl/UnN+GC8Z/TB+2L8af97TunSzBXhkaH6+a5MkrbJpndFnibb6vg7JfmB/N/uNJA9MqZbVtBn40loXMQWOa+MZe2y5bsKVTFZzx6z78x53XE9dSadpBf08cOHQ/Fbg0eEOVXUAODCl/a+JJHNVNbvWdUya49p4Wh2b4xrPtC7d/DuwI8n2JGcCe4FDU9qXJGkZUzmjr6oTSX4X+BfgDOAdVXXvNPYlSVre1G6YqqoPAh+c1vbXqaYuRQ1xXBtPq2NzXGNIVY3uJUnasHzWjSQ1zqA/hVGPcEhyVpIbu+WHk2wbWnZ11/5AkhcOtR9LcneSu5LMrc5IfqDuscaV5LwkH03yjSRvPWmdn+/GdTTJW5Is9fXaqZvS2G7vtnlX9zp/dUbzfTWMO65fS3KkOzZHkjx/aJ01P2ZTGteaH6+ujnHHdvFQ7Z9K8usr3eayqsrXSS8GHyB/DrgIOBP4FLDzpD6/A/xdN70XuLGb3tn1PwvY3m3njG7ZMWDzBh3XOcAvA68B3nrSOncAz2Zw/8SHgMsaGtvtwOwGPWbPBJ7STT8D+MJ6OWZTHNeaHq8JjO3xwKZu+gLgOIPPUkduc7mXZ/RL+94jHKrqO8DiIxyG7QYOdtM3A7u6s6LdwA1V9e2q+jxwtNveejD2uKrqm1X1ceC/hzsnuQB4QlX9Ww1+Ot8N7JnqKJY28bGtE33GdWdVLd6/ci9wdncmuR6O2cTHtSpVr0yfsX2rqk507Wfz/zearmSbp2TQL20lj3D4Xp/uwHwNOG/EugV8uPt1cz+rr8+4ltvm/IhtroZpjG3RO7tfpf98DS5xTGpcvwHcWVXfZn0cs2mMa9FaHi/oObYkv5jkXuBu4DXd8l6PlTHolzbyEQ7L9Flu3edU1bMYPNXzyiTPHb/EsfQZV59troZpjA3gFVX1M8CvdK9XjlFbH73HleTpwHXAb5/GNqdtGuOCtT9e0HNsVXW4qp4O/AJwdZKzV7jNUzLolzbyEQ7DfZJsAn4C+PJy6y7+ullVx4H3s/qXdPqMa7ltbh2xzdUwjbFRVV/o3r8O/D0b7Jgl2crgZ+1VVfW5of5rfcymMa71cLxgQj+LVXU/8E0Gn0OsZJunZNAvbSWPcDgE7OumXwp8pLveeQjY210L3Q7sAO5Ick6SHwdIcg7wAuCeVRjLsD7jWlJVPQZ8Pckl3a/JrwJumXzpI018bEk2JdncTT8OeDEb6JglORf4AHB1Vf3rYud1cswmPq51cryg39i2d8FPkqcCT2PwJY5+j5VZy0+n1/MLeBHwWQafdL++a/sL4CXd9NnAPzD4sPUO4KKhdV/frfcA3bcZGHxa/qnude/iNjfYuI4xOOv4BoMzjJ1d+yyDv1CfA95KdyPeRh8bg2/jHAE+3R2zN9N9g2ojjAv4MwZnhHcNvc5fL8ds0uNaL8er59he2dV+F/BJYM9y21zpyztjJalxXrqRpMYZ9JLUOINekhpn0EtS4wx6SWqcQS9JjTPoJalxBr0kNe7/AF9J4UMQNWATAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x21a91bc6f60>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "VBsB0mv4yC7V",
        "colab_type": "text"
      },
      "source": [
        "### Model for entire data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5l-_TccqyC7W",
        "colab_type": "code",
        "colab": {},
        "outputId": "f54bb386-afb5-4f67-9c84-cf3253efe36c"
      },
      "source": [
        "# load entire dataset for these features. \n",
        "# note where the feature indices are split so we can load the correct ones straight from read_csv\n",
        "n_date_features = 1156\n",
        "X = np.concatenate([\n",
        "    pd.read_csv(\"train_date.csv\", index_col=0, dtype=np.float32,\n",
        "                usecols=np.concatenate([[0], important_indices[important_indices < n_date_features] + 1])).values,\n",
        "    pd.read_csv(\"train_numeric.csv\", index_col=0, dtype=np.float32,\n",
        "                usecols=np.concatenate([[0], important_indices[important_indices >= n_date_features] + 1 - 1156])).values\n",
        "                ], axis=1)\n",
        "y = pd.read_csv(\"train_numeric.csv\", index_col=0, dtype=np.float32, usecols=[0,969]).values.ravel()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\numpy\\lib\\arraysetops.py:472: FutureWarning: elementwise comparison failed; returning scalar instead, but in the future will perform elementwise comparison\n",
            "  mask |= (ar1 == a)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TJWPBRIlyC7Z",
        "colab_type": "code",
        "colab": {},
        "outputId": "fc5ac4b5-c510-4876-dde1-6e3ee237f15c"
      },
      "source": [
        "clf_full_data = XGBClassifier(max_depth=5, base_score=0.005)\n",
        "skf = StratifiedKFold()\n",
        "skf.get_n_splits(X, y)\n",
        "preds = np.ones(y.shape[0])\n",
        "for train_index, test_index in skf.split(X, y):\n",
        "    print(\"TRAIN:\", train_index, \"TEST:\", test_index)\n",
        "    #X_train, X_test = X[train_index], X[test_index]\n",
        "    #y_train, y_test = y[train_index], y[test_index]\n",
        "    \n",
        "    preds[test_index] = clf_full_data.fit(X[train_index], y[train_index]).predict_proba(X[test_index])[:,1]\n",
        "    print(\"ROC AUC: \", roc_auc_score(y[test_index], preds[test_index]))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "TRAIN: [ 394571  394572  394573 ... 1183744 1183745 1183746] TEST: [     0      1      2 ... 395664 395665 395695]\n",
            "ROC AUC:  0.7192622140871269\n",
            "TRAIN: [      0       1       2 ... 1183744 1183745 1183746] TEST: [394571 394572 394573 ... 790831 790909 790942]\n",
            "ROC AUC:  0.7020203750588716\n",
            "TRAIN: [     0      1      2 ... 790831 790909 790942] TEST: [ 789150  789151  789152 ... 1183744 1183745 1183746]\n",
            "ROC AUC:  0.6941438910222872\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "rFxW0yVMyC7h",
        "colab_type": "code",
        "colab": {},
        "outputId": "3f5ceb42-5518-44ff-f4a4-2a5092bff941"
      },
      "source": [
        "print(roc_auc_score(y, preds))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "0.7047588696721703\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QBRUmpvSyC7m",
        "colab_type": "text"
      },
      "source": [
        "### Save Model - Full Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mI5w-EfkyC7m",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# save the model to disk\n",
        "filename = 'XGBoost_Model_Full_Data.sav'\n",
        "pickle.dump(clf_full_data, open(filename, 'wb'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xeS-R2OxyC7r",
        "colab_type": "text"
      },
      "source": [
        "### Load Model - Full Data"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8FXD9NbIyC7s",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# load the model from disk\n",
        "clf_full_data = pickle.load(open(filename, 'rb'))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DMEw0n0ayC7w",
        "colab_type": "code",
        "colab": {},
        "outputId": "e0ba6807-2740-4bf5-a5b3-4fc3f7d31c98"
      },
      "source": [
        "# pick the best threshold out-of-fold\n",
        "thresholds = np.linspace(0.01, 0.99, 50)\n",
        "mcc = np.array([matthews_corrcoef(y, preds>thr) for thr in thresholds])\n",
        "plt.plot(thresholds, mcc)\n",
        "best_threshold = thresholds[mcc.argmax()]\n",
        "print(mcc.max())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "C:\\ProgramData\\Anaconda3\\lib\\site-packages\\sklearn\\metrics\\classification.py:538: RuntimeWarning: invalid value encountered in double_scalars\n",
            "  mcc = cov_ytyp / np.sqrt(cov_ytyt * cov_ypyp)\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "0.19472164446301074\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x21a9260a080>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vWUrKf_xyC7z",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.metrics import confusion_matrix, roc_curve, auc"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "LbLpJZ5-yC74",
        "colab_type": "code",
        "colab": {},
        "outputId": "a3d24079-7e3c-4ac5-f4ca-848518b8293a"
      },
      "source": [
        "confusion_matrix(y, preds>0.08)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "array([[1173116,    3752],\n",
              "       [   5738,    1141]], dtype=int64)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 91
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Z8nHndFOyC76",
        "colab_type": "code",
        "colab": {},
        "outputId": "608ef07a-0dcb-4cc2-924c-746a90419b0a"
      },
      "source": [
        "y.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1183747,)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 50
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YzjLhUt_yC78",
        "colab_type": "code",
        "colab": {},
        "outputId": "0209cbac-db7e-4153-82ab-1b1c1ced907a"
      },
      "source": [
        "y.sum()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "6879.0"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 57
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0Byoc0s0yC7_",
        "colab_type": "code",
        "colab": {},
        "outputId": "d84de24d-7a42-4f25-ad7b-5ff88c348496"
      },
      "source": [
        "preds[preds>.02].shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(9665,)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ymAPb8m3yC8C",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "fpr, tpr, threshold = roc_curve(y, preds)\n",
        "roc_auc = auc(fpr, tpr)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "3xFtwb73yC8D",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "%config InlineBackend.figure_format = 'retina'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "86jcgWdsyC8F",
        "colab_type": "code",
        "colab": {},
        "outputId": "e9810a11-9fe1-4072-ce4f-bb7500695846"
      },
      "source": [
        "plt.figure()\n",
        "plt.plot(fpr, tpr, color='darkorange',\n",
        "         lw=2, label='ROC curve (area = %0.2f)' % roc_auc)\n",
        "plt.plot([0, 1], [0, 1], color='navy', lw=2, linestyle='--')\n",
        "plt.plot(fpr, threshold, color='red', lw=2, linestyle='--')\n",
        "plt.xlim([0.0, 1.0])\n",
        "plt.ylim([0.0, 1.05])\n",
        "plt.xlabel('False Positive Rate')\n",
        "plt.ylabel('True Positive Rate')\n",
        "plt.title('Receiver operating characteristic (ROC)')\n",
        "plt.legend(loc=\"lower right\")\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x21a91c60fd0>"
            ]
          },
          "metadata": {
            "tags": [],
            "image/png": {
              "height": 277,
              "width": 393
            }
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JDh62l9uyC8H",
        "colab_type": "code",
        "colab": {},
        "outputId": "064f1451-c96b-42d5-fadb-c1b3ba134612"
      },
      "source": [
        "plt.plot(fpr, threshold, color='red', lw=2, linestyle='--')\n",
        "plt.ylim([0.0, 0.02])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(0.0, 0.02)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 74
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAxIAAAH4CAYAAADehNsGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAAWJQAAFiUBSVIk8AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvNQv5yAAAIABJREFUeJzs3XmUnFWd//H3NwkJISaBAGGLrAKJIoQkEEIEExgBRQV/4uAyiLiBgCCIoqhs8wNnRtzYBtxAQUV+zCiKCowGEAMJ0JFNgbCFxSCRhAGykITk/v64T9td1VXdVenqru7O+3XOc55+tvvc4nCgPnW3SCkhSZIkSfUY1OwKSJIkSep/DBKSJEmS6maQkCRJklQ3g4QkSZKkuhkkJEmSJNXNICFJkiSpbgYJSZIkSXUzSEiSJEmqm0FCkiRJUt0MEpIkSZLqZpCQJEmSVDeDhCRJkqS6GSQkSZIk1c0gIUmSJKluDQsSETEuIn4QEQsjYmVELIiIb0XEJnWWM6Z4bkFRzsKi3HEV7t00Ij4eET+PiMciYkVEvBQRf4yIj0VE1c8XEftGxG8iYklELI+I+yPiMxExuJNn3hkRtxbvWBoRcyPi6Ho+nyRJkjQQREqp+4VE7ATcAYwFrgceBvYGZgKPANNTSotrKGfTopxdgFnA3cB44DBgETAtpfREu/uPA/4TeA64BXga2AL4P8Bo4L+A96WyDxkRhxXXXgV+BiwB3gXsClyXUnpfhbqdCFwELC6eWQUcAYwDvp5SOq2rzydJkiQNFI0KEjcBBwEnpZQuanf+G8ApwOUppeNqKOdy4JPAN1NKp7Y7fxLwbeCmlNIh7c4fAIwAfp1SWtvu/JbAXcDrgSNSSv/V7too4DFy0JieUrqnOL8hObxMAz6QUrqm3TPbk8PRMmBySmlBcX4TctjZCdg3pXRnV59RkiRJGgi63bUpInYkh4gFwCVll88if/k+KiJGdFHOCOCo4v6zyi5fXJR/cPE+AFJKs1JKv2ofIorzfwMuKw5nlJV1BLA5cE1riCieeRX4cnH4qbJnPgoMAy5uDRHFMy8C5xeHXQYlSZIkaaBoxBiJA4r9zRW+0L8CzAY2AvbpopxpwHBgdvFc+3LWAjcXhzNrrNfqYv9alfreWOGZPwDLgX0jYliNz/y27B5JkiRpwGtEkNi12M+vcv3RYr9LL5VDRAwBPlwcln/5r/qelNJrwJPAEGDHGp95jtyKMi4iNuqqbpIkSdJAMKQBZYwu9i9Vud56fuNeKgfg34DdgN+klG5qwHtqeWZEcd/yzioWES1VLu0GLCV34ZIkSZJ6yvbAyymlHbpTSCOCRFei2Hd3VHdN5RQDsz9LHhx9VE+9pwHPlBs8fPjwMRMmTBjTjTK6Z9kyeOSRtuMRI2DXXavfL0mSpH7noYceYsWKFd0upxFBovVX+tFVro8qu6/HyomIE8izO/0FODCltKRB73kJ2Kx4ptI0tq3PvFytbq1SSpMrnY+IlgkTJkxqaanWYNEL5s6FfdoNZXnjG+Guu5pXH0mSJDXc5MmTmTdv3oLultOIMRKtP2FXG7uwc7GvNvahIeVExGfIszs9CMwsZm6q6z3F2IodyAO0n6jxma3I3ZqeTSl12q2pzxtcthbfmjXNqYckSZL6vEYEiVuK/UHlK0lHxEhgOrACmNNFOXOK+6YXz7UvZxB5itn272t//XTgm8C95BCxqJP3zCr2h1S4tj95hqk7Ukora3zm7WX39F8GCUmSJNWo20EipfQ4eWrW7YETyi6fQ/61/kcppWWtJyNifESMLytnKXBVcf/ZZeWcWJR/U/uVrYuyvkIeXN1C7s70QhdVvg54AXh/RExpV86GwP8tDv+z7JkrgJXAicXidK3PbAKcURxeRn9nkJAkSVKNGjXY+njgDuDCiDgQeAiYSl7zYT7wpbL7Hyr2UXb+DPICcqdGxETy6tQTgMOARZQFlYg4GjgXWAPcDpwUUV4kC1JKV7YepJRejohPkAPFrRFxDbAEeDd5mtfrgJ+1LyCl9GREfA64ELgnIn4GrCIvbjcO+PqAWNXaICFJkqQaNSRIpJQeL37dP5fc/ecdwHPkL97nVBn0XKmcxRExjbyy9eHAfuTBzVcAZ6aUni17pHXKqsHAZ6oUextwZdl7fhERbyUHnPcCGwKPAacCF6aUOsy+lFK6KCIWAKeR16gYRB7U/eWU0g9r+Xx9XnmQWLu28n2SJEla7zVs+teU0jPAMTXe26HZoN21JcDJxdZVOWfTsRtUTVJKs8mBp55nfgX8al3e1y8MKuvpZouEJEmSqmjEYGsNFHZtkiRJUo0MEmpjkJAkSVKNDBJqY5CQJElSjQwSamOQkCRJUo0MEmrTPkhEdBx8LUmSJBUaNmuTBoCxY+G113KA6LgehyRJkvQPBgm1iejYvUmSJEmqwL4rkiRJkupmkJAkSZJUN4OEJEmSpLoZJCRJkiTVzcHWapMSHH98Xj9izRpYuxauuKLZtZIkSVIfZJBQqcsuKz3+wQ+cClaSJEkd2LVJbSotQufq1pIkSarAIKFS5etIGCQkSZJUgUFCpQwSkiRJqoFBQqUMEpIkSaqBQUKlysdIrF3bnHpIkiSpTzNIqJQtEpIkSaqBQUKlDBKSJEmqgUFCpQwSkiRJqoFBQqUMEpIkSaqBQUKlDBKSJEmqgUFCpQwSkiRJqoFBQqWGDCk9NkhIkiSpAoOESpUHidWrm1MPSZIk9WlDur5F65VrrsnhYYMNcqjYccdm10iSJEl9kEFCpSZObHYNJEmS1A/YtUmSJElS3QwSkiRJkupmkJAkSZJUN4OEJEmSpLo52Fqlbr4Znnoqz9z02mvwrnfBDjs0u1aSJEnqYwwSKnXhhfDrX7cd77CDQUKSJEkd2LVJpTbYoPT4tdeaUw9JkiT1aQYJlXJla0mSJNXAIKFStkhIkiSpBgYJlbJFQpIkSTUwSKiULRKSJEmqgUFCpWyRkCRJUg0aFiQiYlxE/CAiFkbEyohYEBHfiohN6ixnTPHcgqKchUW546rcf0REXBQRt0fEyxGRIuLqTsq/srins+33Zc98pIv7j6vnM/ZptkhIkiSpBg1ZRyIidgLuAMYC1wMPA3sDJwOHRMT0lNLiGsrZtChnF2AWcA0wHjgGODQipqWUnih77MvAHsBS4Nni/s78AlhQ5dpRwI7Ab6tcvx64t8L5e7p4Z/9hi4QkSZJq0KgF6S4lh4iTUkoXtZ6MiG8ApwDnAbX8an8+OUR8M6V0artyTgK+XbznkLJnTiEHiMeAtwK3dPaClNIvyGGiRERsDHweWAVcWeXxX6SUql0bGGyRkCRJUg263bUpInYEDiL/yn9J2eWzgGXAURExootyRpBbBJYVz7V3cVH+wcX7/iGldEtK6dGUUlrXz1A4ChgO/HdK6YVultV/2SIhSZKkGjRijMQBxf7mlNLa9hdSSq8As4GNgH26KGca+Yv87OK59uWsBW4uDmd2u8aVfaLYf6eTeyZGxGci4gsRcVS1cRv9mi0SkiRJqkEjujbtWuznV7n+KLnFYhfg91XuqbUcinIaKiKmAW8G5qeUOusadXLZ8ZqI+B7wmZTSqzW+q6XKpa7GdvQOg4QkSZJq0IgWidHF/qUq11vPb9xL5ayLTxb771a5/iTwaXLYGQFsDfwzubvVscAPeqBOzWHXJkmSJNWgUYOtOxPFvrtjGBpVTmmhEaPJoaDqIOuU0m3Abe1OLQf+X0TMAe4DPhAR/55Suq+r96WUJlepRwswqb7a94CDD4aNN86BYoMN4M1vbnaNJEmS1Ac1Iki0thSMrnJ9VNl9PV1Ovf6FPIbjmnoHWaeUnomI3wAfAvYnh4r+bdKkvEmSJEmdaETXpkeKfbWxCzsX+2pjHxpdTr1aB1lfvo7P/73YdzorlSRJkjSQNCJItA5OPigiSsqLiJHAdGAFMKeLcuYU900vnmtfziDygO327+u2iJhKXsxufkrp1nUsZmqxL18oT5IkSRqwuh0kUkqPk6dm3R44oezyOeRf6n+UUlrWejIixkdEySxFKaWlwFXF/WeXlXNiUf5NFVa27o7WQdadTflKROxX4VxExBfJ09a+ANzYwHpJkiRJfVqjBlsfD9wBXBgRBwIPkX+pn0nuivSlsvsfKvZRdv4MYAZwakRMBO4CJgCHAYvoGFSIiMOBw4vDLYv9tIi4svj7hZTSaRWeGwUcSR5k/cMuPt8fImI+cDfwV/I4junAbuSB1x9KKb3cRRmSJEnSgNGQIJFSejwipgDnAocA7wCeAy4EzkkpLamxnMXFmg5nkcPBfsBi4ArgzJTSsxUemwgcXXZux2IDeAroECTIA6RHUNsg6wuAvcmL740B1gJPk1fy/kaDW0ma63//F558Mq8fsXo1bLIJTJjQ7FpJkiSpj4mUGjqbqtZRRLRMmjRpUktLtfXqesm118KRR7Ydv/e9cN11zauPJEmSGmry5MnMmzdvXrVlCWrViMHWGkiGDSs9dkE6SZIkVWCQUKmhQ0uPV65sTj0kSZLUpxkkVKq8RcIgIUmSpAoMEipVHiRWrWpOPSRJktSnGSRUyq5NkiRJqoFBQqXs2iRJkqQaGCRUyq5NkiRJqoFBQqXs2iRJkqQaGCRUyq5NkiRJqoFBQqUMEpIkSaqBQUKlyrs2OUZCkiRJFRgkVMoWCUmSJNXAIKFSG2xQevzaa7B2bXPqIkmSpD5rSLMroD4mAj76URg8OLdODBuWg8QgM6ckSZLaGCTU0fe/3+waSJIkqY/zZ2ZJkiRJdTNISJIkSaqbQUKSJElS3QwSkiRJkupmkFBlKeXF6F55JU8BK0mSJLVjkFBHb3lLnu512DAYNQruuafZNZIkSVIfY5BQRxGlx6tWNacekiRJ6rMMEupo2LDS45Urm1MPSZIk9VkGCXU0dGjpsUFCkiRJZQwS6qi8RcKuTZIkSSpjkFBHdm2SJElSFwwS6sggIUmSpC4YJNRR+RgJuzZJkiSpjEFCHdkiIUmSpC4YJNSRg60lSZLUBYOEOnL6V0mSJHXBIKGO7NokSZKkLhgk1NHw4aXHK1Y0px6SJEnqswwS6mijjUqPly9vTj0kSZLUZw1pdgXUB221Fey2G4wYkUPFG97Q7BpJkiSpjzFIqKMjjsibJEmSVIVdmyRJkiTVzSAhSZIkqW4GCUmSJEl1a1iQiIhxEfGDiFgYESsjYkFEfCsiNqmznDHFcwuKchYW5Y6rcv8REXFRRNweES9HRIqIqzspf/vinmrbNZ08e3RE3BURSyPipYi4NSLeWc/nkyRJkgaChgy2joidgDuAscD1wMPA3sDJwCERMT2ltLiGcjYtytkFmAVcA4wHjgEOjYhpKaUnyh77MrAHsBR4tri/FvcBv6hw/sEqdbsA+Gzxju8CQ4H3A7+KiE+nlC6u8b2SJElSv9eoWZsuJYeIk1JKF7WejIhvAKcA5wHH1VDO+eQQ8c2U0qntyjkJ+HbxnkPKnjmF/OX+MeCtwC011vnelNLZtdwYEfuSQ8TjwF4ppReL818DWoALIuKGlNKCGt/dtz3/PBx+eF6IbvlyGDMG5sxpdq0kSZLUh3S7a1NE7AgcBCwALim7fBawDDgqIkZ0Uc4I4Kji/rPKLl9clH9w8b5/SCndklJ6NKWU1vUz1KA1BJ3XGiKKdy8gf+Zh5FaTgWHQoBwc7rsPHn0UHnus2TWSJElSH9OIMRIHFPubU0pr219IKb0CzAY2AvbpopxpwHBgdvFc+3LWAjcXhzO7XeNs64g4NiLOKPa7d3Jv62e8scK135bd0/+NKMt8S5c2px6SJEnqsxrRtWnXYj+/yvVHyS0WuwC/72Y5FOU0wtuK7R8i4lbg6JTS0+3OjQC2AZamlJ7rbr0ioqXKpVrHdvS84cNzq8TaIheuXAmrV8MGGzS3XpIkSeozGtEiMbrYv1Tleuv5jXupnK4sB/4VmAxsUmytYytmAL8v64bVW/XqOyJg5MjSc6+8UvleSZIkrZcaNdi6M1HsuzuGoSHlpJQWAWeWnf5DRBwE/BGYCnycPLi7rqJrfP/kSueLlopJdb6z54wcCS+1y06vvJIHXUuSJEk0pkWi9dvm6CrXR5Xd19PlrJOU0mvA94rD/euoV1ctFv2TLRKSJEnqRCOCxCPFvtoYgZ2LfbWxD40upzv+Xuz/0bUppbQM+CvwuojYqkn16n2jRpUeGyQkSZLUTiOCROu6DQdFREl5ETESmA6sALpaiGBOcd/04rn25QwiD9hu/76e0DqzVPmid7OKffkaFgBvL7tnYChvkXj55ebUQ5IkSX1St4NESulx8tSs2wMnlF0+h/zr/o+KX/YBiIjxEVEyS1FKaSlwVXH/2WXlnFiUf1OFla3rEhFTI2JohfMHkBe3A7i67PJlxf5LEbFJu2e2J3/mlcAV3alXn2PXJkmSJHWiUYOtjwfuAC6MiAOBh8iDlmeSu/x8qez+h4p9lJ0/gzxz0qkRMRG4C5gAHAYsomNQISIOBw4vDrcs9tMi4sri7xdSSqe1e+TfgTcVU70+W5zbnbZ1IL6SUrqj/TtSSncUq3SfCtwfEdcBQ4EjgTHApwfMqtatDBKSJEnqREOCRErp8YiYApxL7v7zDuA54ELgnJTSkhrLWRwR08grWx8O7AcsJv/af2ZK6dkKj00Eji47t2OxATwFtA8SVwHvAfYid0vaAHgeuBa4OKV0e5W6fTYi7ie3jnwSWAvMA76WUrqhls/XrxgkJEmS1ImGTf+aUnoGOKbGe8tbItpfWwKcXGy1lHU2HbtCdXb/94Hv13p/2bM/BH64Ls/2OwYJSZIkdaIRg601EDlrkyRJkjrRGwvSqT9qbZEYPjyHimHDmlsfSZIk9SkGCVV27LFw3HEwxH9FJEmS1JHfElXZBhs0uwaSJEnqwxwjIUmSJKluBglJkiRJdTNISJIkSaqbQUKSJElS3RxsrcpWroRTT83rR7zyCqxZA7/8ZbNrJUmSpD7CIKHKBg+GSy9tO46AlPJekiRJ6z27NqmyIUPyYnStUoJly5pXH0mSJPUpBglV17q6dauXX25OPSRJktTnGCRUXXmQeOWV5tRDkiRJfY5BQtWNGlV6bJCQJElSwSCh6myRkCRJUhUGCVVnkJAkSVIVBglVZ5CQJElSFQYJVWeQkCRJUhUGCVU3enTp8f/+b3PqIUmSpD7HIKHqxowpPX7xxebUQ5IkSX2OQULVlQeJJUuaUw9JkiT1OQYJVWeQkCRJUhUGCVVnkJAkSVIVQ5pdAfVhU6bA3Lk5UIwZ03HwtSRJktZbBglVN3o07L13s2shSZKkPsiuTZIkSZLqZpCQJEmSVDeDhCRJkqS6GSQkSZIk1c3B1urcrFnwwAOwaBH8/e/wqU/Bnns2u1aSJElqMoOEOnf55XDttW3H++9vkJAkSZJdm9SFzTcvPV60qDn1kCRJUp9ikFDnxo4tPTZISJIkCYOEurLFFqXHzz/fnHpIkiSpTzFIqHMGCUmSJFVgkFDnDBKSJEmqwCChzhkkJEmSVIFBQp0rDxKLFkFKzamLJEmS+gyDhDo3YkTeWq1eDS++2Lz6SJIkqU9oWJCIiHER8YOIWBgRKyNiQUR8KyI2qbOcMcVzC4pyFhbljqty/xERcVFE3B4RL0dEioirOyl/54g4PSJmRcQzEbEqIp6PiOsjYmaVZz5SlFttO66ez9jv2L1JkiRJZRqysnVE7ATcAYwFrgceBvYGTgYOiYjpKaXFNZSzaVHOLsAs4BpgPHAMcGhETEspPVH22JeBPYClwLPF/Z35V+BI4C/Ab4AlwK7Au4F3R8TJKaULqzx7PXBvhfP3dPHO/m3LLeGJdv/Yn38eJkxoXn0kSZLUdA0JEsCl5BBxUkrpotaTEfEN4BTgPKCWX+3PJ4eIb6aUTm1XzknAt4v3HFL2zCnkAPEY8Fbgli7ecSPw7ymlP7U/GRFvBf4H+FpE/L+U0nMVnv1FSunKGj7HwFLeIvG3vzWnHpIkSeozut21KSJ2BA4CFgCXlF0+C1gGHBURI+hEcf2o4v6zyi5fXJR/cPG+f0gp3ZJSejSl2kYAp5SuLA8RxfnbgFuBocC+tZS13thmm9LjZ59tTj0kSZLUZzRijMQBxf7mlNLa9hdSSq8As4GNgH26KGcaMByYXTzXvpy1wM3FYcVxDA2yuti/VuX6xIj4TER8ISKOqjZuY8B5/etLj595pjn1kCRJUp/RiK5Nuxb7+VWuP0pusdgF+H03y6Eop+EiYjvgQGA58Icqt51cdrwmIr4HfCal9GpP1KtPOPRQ2GorGDcuh4px60d+kiRJUnWNCBKji/1LVa63nt+4l8qpW0QMA34MDAM+n1Iqn9/0SeDT5FaRZ8l1fQvwVeBYYBTwwRrf1VLlUleDxJvnTW/KmyRJklTojXUkoth3dxWzRpVTWmjEYOAqYDrwM+CC8ntSSrellC5OKc1PKS1PKT2XUvp/5G5WLwIfiIg9GlkvSZIkqS9rRItEa0vB6CrXR5Xd19Pl1KwIEVcD7wOuBf6l1kHbACmlZyLiN8CHgP2B+2p4ZnKVurQAk2p9tyRJktRMjWiReKTYVxu7sHOxrzb2odHl1CQihgA/Bd4P/AT4YEqp2iDrzvy92Hc6K5UkSZI0kDQiSLSu23BQRJSUFxEjyV2GVgBzuihnTnHf9OK59uUMIg/Ybv++dRYRQ4HryC0RPwKOSimtWcfiphb78oXyJEmSpAGr20EipfQ4eRDy9sAJZZfPIf9S/6OU0rLWkxExPiJKBhenlJaSxyqMAM4uK+fEovybKqxsXZdiYPXPgcOA7wPHlE9bW+GZ/Sqci4j4Inna2hfIC90NXI8/DtdfDxdfDKefDrNnN7tGkiRJaqJGrWx9PHAHcGFEHAg8RP6lfia5K9KXyu5/qNhH2fkzgBnAqRExEbgLmED+0r+IjkGFiDgcOLw43LLYT4uIK4u/X0gpndbukcuAd5C//P8VODOivBrcmlK6td3xHyJiPnB38cxockvLbuTpYj+UUnq5vJAB5aKL4NvfbjseMwamT29efSRJktRUDQkSKaXHI2IKcC5wCPmL+nPAhcA5KaUlNZazOCKmkVe2PhzYD1gMXAGcmVKqtKTyRODosnM7FhvAU0D7ILFDsd8MOLOT6tza7u8LgL3Ji++NAdYCT5NX8v5Gd1tJ+oXytSNclE6SJGm91qgWCVJKzwDH1HhvhyaAdteWkBd+K1/8rdr9Z9OxK1Rn98+o9d52z3yu3mcGnPLVrZ96qjn1kCRJUp/QG+tIaCDYaafS48cea049JEmS1CcYJFSbN7yh9PiJJ2DNuk50JUmSpP7OIKHabLwxbLZZ2/GqVY6TkCRJWo8ZJFS7nXcuPX700ebUQ5IkSU1nkFDtyoOE4yQkSZLWWwYJ1a58nIQtEpIkSestg4RqZ9cmSZIkFQwSqp1dmyRJklQwSKh2TgErSZKkgkFCtRs9GjbfvO141Sp4+unm1UeSJElNY5BQfXbdtfTYcRKSJEnrpSHNroD6meOOgyOPhIkTYffdYdSoZtdIkiRJTWCQUH0+9KFm10CSJEl9gF2bJEmSJNXNICFJkiSpbgYJSZIkSXUzSEiSJEmqm4OtVb9XXoG774b77oN774WddoIzz2x2rSRJktSLDBKq35w5cNBBbceTJhkkJEmS1jN2bVL99tij9PjBB2H16ubURZIkSU1hkFD9xo6FrbZqO161Ch55pHn1kSRJUq8zSGjdTJxYenz33c2phyRJkprCIKF1s9depcdz5jSnHpIkSWoKg4TWzbRppccGCUmSpPWKQULrZurU0uMHHoCXXmpOXSRJktTrDBJaN5tsAm98Y9txSnDHHc2rjyRJknqVQULrbr/9So9vv7059ZAkSVKvM0ho3ZUHidtua049JEmS1OsMElp3M2aUHs+dCy+/3JSqSJIkqXcZJLTuttkGxo9vO16zxlYJSZKk9YRBQt3zT/9Uevy73zWnHpIkSepVBgl1T2uQiIApU2CXXZpbH0mSJPWKIc2ugPq5mTPhxz+Gt70NNt+82bWRJElSLzFIqHtGjYIPfrDZtZAkSVIvs2uTJEmSpLoZJCRJkiTVzSAhSZIkqW4GCTXe/PlOAytJkjTAOdhajbFkCfzHf8D118PDD8PWW8Mzz8Ags6okSdJA5Lc8Ncbw4XDJJTlEACxcCHfe2dw6SZIkqcc0LEhExLiI+EFELIyIlRGxICK+FRGb1FnOmOK5BUU5C4tyx1W5/4iIuCgibo+IlyMiRcTVNbxn34j4TUQsiYjlEXF/RHwmIgZ38sw7I+LWiHgpIpZGxNyIOLqezzdgDR8O73xn6bmf/rQ5dZEkSVKPa0iQiIidgBbgGOAu4JvAE8DJwJ0RsWmN5WwK3Fk893hRzl1FuS0RsWOFx74MnAhMBP5a43sOA/4A7A/8HLgEGFq875oqz5wI/ArYDbga+C6wNXBlRFxQy3sHvH/+59Ljn/4UVq5sTl0kSZLUoxrVInEpMBY4KaV0eErpCymlA8hfzHcFzquxnPOBXYBvppQOLMo5nBwsxhbvKXdK8cwo4FNdvSAiRpFDwBpgRkrpYymlz5GDyJ3AERHx/rJntgcuAJYAU1JKJ6SUTgF2Jweez0bEtBo/48D1jnfAmDFtx0uWwG9+07z6SJIkqcd0O0gUrQQHAQvIv+y3dxawDDgqIkZ0Uc4I4Kji/rPKLl9clH9weatESumWlNKjKaVUY5WPADYHrkkp3dOunFfJrRvQMZB8FBgGXJxSWtDumRfJ4QfguBrfP3ANGwbvf3/puauuak5dJEmS1KMa0SJxQLG/OaW0tv2FlNIrwGxgI2CfLsqZBgwHZhfPtS9nLXBzcTizQfW9scK1PwDLgX0jYliNz/y27J7121FHlR7fcENumZAkSdKA0oggsWuxn1/l+qPFfpdeKqcrVd+TUnoNeJI8Le6ONT7zHLkVZVxEbNTVyyOipdIGjK/zc/RNU6fCG97Qdrx6NfzkJ82rjyRJknpEI4LE6GL/UpXrrec37qVyurIu76n1mdFVrq8/IuDDHy49d/nlUHPPM0mSJPUHvbGORBT77n6TbFQ5PfGemp9JKU205p7TAAAgAElEQVSutAEP11vRPuuYY0oXonvwQdeUkCRJGmAaESS6+jV+VNl9PV1OV9blPbU+83I36jVwjBvXcU2Jyy9vTl0kSZLUIxoRJB4p9tXGLuxc7KuNfWh0OV2p+p6IGALsALxGXgejlme2AkYAz6aUlnezbgPHsceWHl97Lbz4YnPqIkmSpIZrRJC4pdgfFBEl5UXESGA6sAKY00U5c4r7phfPtS9nEHmK2fbvW1eziv0hFa7tT55h6o6UUvuV1Dp75u1l9wjg4INh223bjl99Fa68smnVkSRJUmN1O0iklB4nT826PXBC2eVzyL/W/yiltKz1ZESMj4iSWYpSSkuBq4r7zy4r58Si/JtSSk/QPdcBLwDvj4gp7eq0IfB/i8P/LHvmCmAlcGKxOF3rM5sAZxSHl3WzXgPL4MHwyU/mvyPg7W+HPfZobp0kSZLUMEMaVM7xwB3AhRFxIPAQMJW85sN84Etl9z9U7KPs/BnADODUiJgI3AVMAA4DFtExqBARhwOHF4dbFvtpEXFl8fcLKaXTWu9PKb0cEZ8gB4pbI+Ia8orV7yZP83od8LP270gpPRkRnwMuBO6JiJ8Bq8iL240Dvp5ScjRxuU98AlasgI9/HLbfvtm1kSRJUgNF7QtCd1FQxOuBc8ndfzYFngN+AZyTUlpSdm8CSCmVBwkiYgx5ZevDga2AxeRF385MKT1b4f6z6bgSdntPpZS2r/DcdHLAmQZsCDwG/AC4MKW0pspnfBdwGjCJ3JrzF/Jq1z/s5P01iYiWSZMmTWppaeluUZIkSVJVkydPZt68efOKmUPXWcOChLrHICFJkqTe0Kgg0RvrSEiSJEkaYAwSao5ly7q+R5IkSX2WQUK965578gxOBx0EdquTJEnqtxo1a5PUuZUr4cgj4frr287ddBMcUmlpDkmSJPV1tkiodwwbBmvKJsP6whc6npMkSVK/YJBQ7zn33NLj++6DH/2oOXWRJElStxgk1Hv23BM+8IHSc2ecAUuXNqc+kiRJWmcGCfWu88/P3Zxa/e1v8G//1rz6SJIkaZ0YJNS7tt8ePvvZ0nNf/zo8/XRTqiNJkqR1Y5BQ7/vCF2DLLduOX30VTj+9efWRJElS3QwS6n0jR8J555Weu+Ya+N3vmlMfSZIk1c0goeY4+ug8+Lq9446D5cubUx9JkiTVxSCh5hg8GC69FCLazj3+OHz+882rkyRJkmpmkFDz7LMPHH986blLLoHf/KY59ZEkSVLNDBJqrq9+FXbaqe140CB46KHm1UeSJEk1MUiouUaOhB//OHd12m47uO22jtPDSpIkqc8Z0uwKSEydCtddBzNmwMYbN7s2kiRJqoFBQn3D4Yc3uwaSJEmqg12bJEmSJNXNIKG+bcUK+NGPml0LSZIklbFrk/quZ56B97wHWlrg1Vfhk59sdo0kSZJUsEVCfdOdd8KUKTlEAJx4Ivzxj82tkyRJkv7BIKG+acQIWLq07Xj1anjve3MrhSRJkprOIKG+affdO46NWLQoz+60fHlz6iRJkqR/MEio73rve+ErXyk9N28efPSjsGZNc+okSZIkwCChvu7ss+Hd7y4997OfwfHHQ0pNqZIkSZIMEurrBg2Cq66CCRNKz3/nO3D66YYJSZKkJjFIqO8bNQp+/WvYZpvS81/7Wu76ZJiQJEnqdQYJ9Q877AC/+x1stlnp+fPOg89/3jAhSZLUywwS6j/Gj4ebbsotFO1dcAF8+tOwdm1z6iVJkrQeMkiof5k0KbdMbLxx6fnbboNXXmlOnSRJktZDBgn1P3vtBbNmwaab5uNddoH/+R8YPbq59ZIkSVqPGCTUP+25Z26FmDYNfv972HLLZtdIkiRpvTKk2RWQ1tmb3gSzZ0NEs2siSZK03rFFQv1bZyHirrvg6ad7ry6SJEnrEYOEBqb58+Ed74B99oE//anZtZEkSRpwDBIaeBYtgre/HRYvhueeg/32g+uvb3atJEmSBhSDhAaeiy+GJ55oO162DN7zHvjqV124TpIkqUEMEhp4zj4bTjml9FxKcMYZ8OEPw4oVTamWJEnSQNKwIBER4yLiBxGxMCJWRsSCiPhWRGxSZzljiucWFOUsLMod14h3R8TZEZG62B4ve2ZGF/f/Wz2fUT1s0CD4xjfg0kth8ODSa1dfDfvuC48+2py6SZIkDRANmf41InYC7gDGAtcDDwN7AycDh0TE9JTS4hrK2bQoZxdgFnANMB44Bjg0IqallJ4oe6bed9/aSRXeBUwCflvl+m1Vnv9jpx9MzfGpT8Guu8IRR8CLL7adv/fevEL2d74DH/hA8+onSZLUjzVqHYlLyV/kT0opXdR6MiK+AZwCnAccV0M555NDxDdTSqe2K+ck4NvFew7pzrtTSrdSIQxExGDgY8Xhd6rU79aU0tk1fA71FQccAHPnwrveBY880nZ+6VL44Afhllvg29+G4cObV0dJkqR+qNtdmyJiR+AgYAFwSdnls4BlwFERMaKLckYARxX3n1V2+eKi/IOL9zX03YV3AOOAOSml+2u4X/3FzjvnMPG+93W89t3vwt57w8MP9369JEmS+rFGjJE4oNjfnFJa2/5CSukVYDawEbBPF+VMA4YDs4vn2pezFri5OJzZA+8G+GSxr9YaAfCGiDgxIs6IiI9GxM41lKu+YPRo+NnP8riJYcNKrz34IEyZAk891Zy6SZIk9UONCBK7Fvv5Va63jmrdpQfKaci7I2Ib4O3AS8DPOrn1Q8BF5O5S3wfmR8R19Q4oV5NE5HETc+bkVor2PvAB2G675tRLkiSpH2pEkBhd7F+qcr31/MY9UE6j3v1xYDBwdUppeYXrfwe+ALwZGAlsTg4efwLeC/wqImr6ZxkRLZU28qBy9YaJE6GlJY+RANhttzxOQpIkSTVr1GDrzkSx7+5KYOtSTpfPFAHgo8VhxW5NKaU/A39ud2opcGNE3AHcC0wnz/jk8sn9xciReSrYAw7I08FutFGzayRJktSvNKJFovVX/9FVro8qu6+R5TTi3W8HtmUdBlmnlF4GflIc7l/jM5MrbeRpa9WbIuBjH4MJEypfX7MGjjkG7rijd+slSZLUDzQiSLTOqVltHEJrZ/Rq4xi6U04j3t06yPryTmtX3d+LfS0zQ6k/ufRSuPJKeMtb4IQT4KWusrAkSdL6oxFB4pZif1D5OIGIGEnu9rMCmNNFOXOK+6YXz7UvZxB5mtf27+v2uyNia+BQcovFtV3Ur5rWGaGe6PQu9S8LFsAXv5j/TimHil13zd2hUnd76UmSJPV/3Q4SKaXHyVOzbg+cUHb5HPIv9T9KKS1rPRkR4yOiZHBxSmkpcFVx/9ll5ZxYlH9T+5Wt1+XdZT5GHmR9VZVB1q31nV5pMHVE/AtwJLCKdQ8i6ot+9StYVvavzfPPw1FH5XEVDzzQnHpJkiT1EZEa8OtqROwE3EFeYfp64CFgKnnNh/nAvimlxe3uTwAppSgrZ9OinF2AWcBdwATgMGBRUc7j3Xl3u+cGkVsRtgN2TylV/WYYEQvIoesO4FlgQ2AvYG/gNeATKaUrO/2H1IWIaJk0adKklpaW7hSjRpo1C449Fh57rOO1QYPyVLLnngtjxvR+3SRJktbR5MmTmTdv3rxinO46a0TXptaWgSnAleQv8Z8FdgIuBKZV+iJfpZzF5IXpLgTeUJQzFbgCmFweIrr57oPJIWJOZyGi8J/k8RjTyS0fHwc2K945pbshQn3UAQfA/ffDV77ScRG7tWvhkkvyehSXXZYHZkuSJK1HGtIioe6zRaKPe/RR+PSn4aabKl/fYw+44AL4p3/q3XpJkiTVqU+1SEgD3s47w29/C9dfDzvu2PH6fffB294G73xnbq2QJEka4AwSUq0i4N3vhj//Gc4/v/Iidm94Qx4/IUmSNMD5jUeq14Yb5qlh58+HD36w7fzIkfClLzWvXpIkSb3IICGtq222gR//GObOhZkz4fTTYfPNK9/74IPw17/2bv0kSZJ6kEFC6q69985TxX7hC5WvpwTHHJPHVpxwAjz3XO/WT5IkqQcYJKRGGTy48vn//m+45x5YtSqvkL3jjnDyybBwYe/WT5IkqYEMElJPeu21juMmXn0VLrwQdtgBPvGJPNZCkiSpnzFISD3t1FNhu+06nl+1Cr73PRg/Ht7zHrj99twNSpIkqR8wSEg9acgQ+OQnc6vDpZfCuHEd70kJfvEL2H9/mDgRLr8cli7t/bpKkiTVwSAh9YahQ+FTn4LHHoP//M/cramS+++H447LM0Kdfnrv1lGSJKkOBgmpNw0bloPC/Pnw05/mFohKXn7Z2Z0kSVKfZpCQmmHIEHj/+2HePPjd7+DwwzuuiH3CCc2pmyRJUg0MElIzRcCBB8LPfw5PPglnnJEXtZs8Oa9PUcmsWfChD8Gddzo4W5IkNY1BQuortt0WzjsPnnkGrr02h4xKLrwQfvIT2HffHDi+9z1Yvrx36ypJktZ7Bgmprxk2LC9aV8nTT8OvftV2/Kc/5bUott4ajj8e5s61lUKSJPUKg4TUn/z0p7B2bcfzL72UZ4PaZ5+8LsX558Ozz/Z+/SRJ0nrDICH1J5/7HNx4I7zrXdW7Ps2fn1fT3nZbeNvb4Ic/hFde6d16SpKkAc8gIfUngwbBwQfDL38JTzyR15oYO7byvSnlGaE+8hHYYgu4+uperaokSRrYDBJSf7X99vBv/5a7MP3853kK2Q02qHzvihUwYUKvVk+SJA1sBgmpv9tggxwifv5zWLgwz+o0aVLpPTvt1PFcqxtugO98BxYt6vm6SpKkAcMgIQ0km20Gn/40tLTAgw/CF7+Yx0oceWT1MRXf/CYceyxstRXMmAEXXeRAbUmS1KVIThXZJ0REy6RJkya1tLQ0uyoaaNauhVdfhY026njthRdgyy1hzZqO16ZOhfe+N2/VpqOVJEn9zuTJk5k3b968lNLk7pRji4Q00A0aVDlEQB60XSlEQF6T4vOfz92i9twT/vVf4S9/6bl6SpKkfsUgIa3PZsyAc8+FiRM7v+/ee+HMM+FNb8qDtr/4xXxOkiSttwwS0vpsxx3hK1/JK2Q/9hj8x3/kLk2defjhPFvUDTf0Th0lSVKfZJCQlO20U17wbs4ceOaZPPvTW9+au0ZVcuihlc8/+yz8+tewdGnP1VWSJDWdQUJSR+PG5dmfbr0VnnsuTw978MEwZEi+vtVW1btDXXstvPOdMGYMvOUtuevU7NmwenWvVV+SJPW8Ic2ugKQ+buxY+MQn8vbii3DjjbBsWfXpZG+8Me9Xr84BYvZsOOsseN3rYP/94cAD89iMPfaAwYN77WNIkqTGMkhIqt0mm8AHPlD9+rJlcNttla8tXQq/+U3eWsuaMQMOOCBvEyZUDyeSJKnPsWuTpMZZtgw+/nHYeeeu733xxbwa96c/nWeD2nZbWLWq5+soSZIawiAhqXHGjoVLLoH58+Hpp+H734cPfhC22KLrZ7fZBoYOrXzt4YdhxYrG1lWSJHWLXZsk9YzXvx4++tG8pQR//jP8/vd5APdtt+UWifZmzKhcTkp5XMXf/w6TJ8P06W3b2LE9/SkkSVIVBglJPS8CdtstbyefnFfTvu8+mDUrh4vbb4eZMys/u2ABLFyY/54zJ29f/3o+3nnn0mCx667Vp6uVJEkNZZCQ1PsGD4ZJk/J22ml5hqdqA61nz65ezqOP5u3KK/Px6NF5Qb29987bPvvA5ps3vPqSJMkgIakv2GCD6tdWr4btt88tE1156SW4+ea8AZx4Ilx0USNqKEmSytgHQFLfdswx8OST8Ne/5sXuTj4ZpkypbQ2KvfeufP755+HYY/Ng8AcfzF2tJElSXWyRkNQ/bL01vO99eYM81ezcuW2L3s2dC//7v6XPTJ1auaw5c/Jq3d/5Tj5+3etyOGntEjV1ap5FynUtJEmqyiAhqX8aMaJtMTuAtWvhkUfg7rtzqPjLX+ANb6j87F13lR4vXZpnk7r11rZzm20Ge+6ZA8aUKXnGqG23NVxIklQwSEgaGAYNyqtjT5gAH/5w5/fOndt1eS+8AP/zP3lrtdlmcPXVcPDB3aurJEkDgEFC0vrnS1+C/ffPLRNz5+bQUIsXXshdrCp55BF4/PHcclHLAnySJPVzDRtsHRHjIuIHEbEwIlZGxIKI+FZEbFJnOWOK5xYU5Swsyh3XqHdHROpkm9PJe94ZEbdGxEsRsTQi5kbE0fV8Pkl9wMyZcOaZcMMNsGgRPPEEXHMNnHJKXo9i+PDKzw0fnls8KvnJT+DQQ2HLLWHcODjsMDj3XPj1r+Fvf+u5zyJJUpM0pEUiInYC7gDGAtcDDwN7AycDh0TE9JTS4hrK2bQoZxdgFnANMB44Bjg0IqallJ5o0LufAq6scP7ZKnU7EbgIWAxcDawCjgCujIg3p5RO6+rzSeqDImCHHfJ25JH53Jo1eX2Klha45568/elPsPvuMKTKfzbnzWv7+69/zdsvf9l2bqutYI89chl77AETJ8Iuu1QvT5KkPi5SSt0vJOIm4CDgpJTSRe3OfwM4Bbg8pXRcDeVcDnwS+GZK6dR2508Cvg3clFI6pLvvjogE3JZSmlHj59ueHFCWAZNTSguK85sAdwM7AfumlO6spbwq72iZNGnSpJaWlnUtQlJPWrMGFi+GsWMrX996a3juufrKHD48B4prr82tGJIk9YLJkyczb968eSmlyd0pp9tdmyJiR/IX+QXAJWWXzyJ/+T4qIkZ0Uc4I4Kji/rPKLl9clH9w8b6GvrsGHwWGARe3hgiAlNKLwPnFYZdBSVI/Nnhw9RCxahW8+915dqehQ2svc8WK3OpRbfXtOXPyTFIvvlh3dSVJ6mmNaFMv5l7k5pTS2vYXUkqvRMRs8pf9fYDfd1LONGB4Uc4rZeWsjYibya0VM4HW7k3deffGEfFRYEvgJaAlpVRtfETre26scO23ZfdIWt8MHQqXXZb/XrUK/vznHBBaWnKXp/vug5UrKz+7++4wbFjla+efD7/6Vf57u+3yQO7222abNf6zSJJUo0YEiV2L/fwq1x8lf5nfhc6DRC3lUJTTiHfvAXy//YmIuA84KqX0QK11Syk9FxHLgHERsVFKaXmVurS+o1rfpfGdPSepnxg6NK8/seee8PGP53OrV+dZne6/P2/33pu355/PrRjV3Htv299PPZW3//7vtnPjxuWuUZMmwZvfDLvtBjvvXNuq35IkdVMjgsToYv9Sleut5zfugXLW9d3fAP6LHAxeJX+JP508eHpWRExMKf21zveMKO7rNEhIWg9tsEH+kr/bbvDBD7adX7gQXnut8jNLlsAzz3Re7rPP5u2GG9rOjRgBe+0Fv/99XltDkqQe0hvThbQuA9vdUd3rUk7FZ1JKny277x7gfRFxHfBe4DTyQO2G163aoJaipWJSHe+U1N9VW5MCYPly+Jd/ya0SDz9cPXCUW7Ysh5BqIeKKK3KrycSJsOuuzholSVpnjfg/SOuv9KOrXB9Vdl8jy2nUu1tdRg4S+1eo22bFeypNJdv6npdrfI8kdW7cOLjqqvz3q6/CAw+0jbtoaYEHH8xdpirZa6/q5Z59Njz9dP57ww1zl6iJE2GnnfJCenvuCW96kwFDktSlRvyf4pFiv0uV6zsX+2rjGLpTTqPe3ervxb58lqdHyEFiF6BkiteI2Kq4/9muxkdI0jrZcMMcDtoHhFWr4C9/yYO5778/B4377stT1FYbd7F4cVuIgBxQ7r47b+0NH57Xupg8OY+/mDQJ3vjG+makkiQNeI0IErcU+4MiYlD72ZMiYiQwHVgBVF0xujCnuG96RIxsP3NTRAwiD5pu/75GvrvVPsX+ibLzs4qyDqEsSABvb3ePJPWO1u5JEye2nUspB4WRIys/037wdmdWrMhTz85p95/OKVM6Bg5J0nqt2yPxUkqPAzcD2wMnlF0+h/xr/Y9SSstaT0bE+IgomaUopbQUuKq4/+yyck4syr+p/crW6/juSZXWlYiI3YHzisOryy5fAawETiwWp2t9ZhPgjOLwsvIyJalXReRpYseMqXx9m23gC1+AQw7J3Zjqsfvu1a+ddhpccgncdVf1aW4lSQNOozrBHg/cAVwYEQcCDwFTyWs+zAe+VHb/Q8U+ys6fAcwATo2IicBdwATgMGARHcPCurz7JOD/RMQs4BlyQBhPbm0YDHwX+Gn7B1JKT0bE54ALgXsi4mfAKvIsT+OAr3dnVWtJ6hXjx8NXv9p2/Le/5VaK++/PU9E++mgOA88/3/HZyVUWP33xRfj619uON9igrVtUa4vJHnvk7lKSpAGlIUEipfR4REwBziV/IX8H8Bz5i/c5KaUlNZazOCKmkVelPhzYjzy4+QrgzJTSsw149y/Ig6N3Jy8it2Hxjt8C300p/bJK3S6KiAXkGZ0+TG7N+Qvw5ZTSD2v5fJLUp2y5ZW6dOOSQtnMp5Wlp581r21pa8jiJSv70p9Lj1avhnnvy1mrw4Dz17ZQpeZzH1Kn52AHdktSvRUrdnZVVjRARLZMmTZrU0lJtvTpJaqKUctepchdcAJ/7XP3lbbQRnHoq/Ou/dr9ukqS6TJ48mXnz5s2rtixBrfw5SJLUtUohAuDgg2HNmjwQ+667ul5Er9Xy5bDJJpWvvfoq/PGPuQVj467WMpUkNYtBQpK07t785ry1ev753K3p3nvz1tICTz5Z+dmpUyufnzcP3va2/PfOO7dNfbvXXnmdi402auxnkCStE4OEJKlxttgCDj00b62WLGkbNzF3bp5WdsmS6uMu5s5t+/vRR/P2k5/k4yFD8gDuadPatu22q95iIknqMQYJSVLPGjMGDjoob5DHWyxaVH0mp/ZBotxrr7WFkosuyue22CIHiunT4YgjYPvtG1p9SVJl3V5HQpKkukR0vo7FjjvmWZ0G1fi/qOefh1/8Ig/6fuihru+XJDWEQUKS1Lecfz488AC8/DL84Q95nYr3vz8HjK5Mm1b5/M9/Dh/5CHznO/Dgg7B2bUOrLEnrI7s2SZL6phEjYL/98tZq0SK488627e67YcWKfO1Nb6o+y9Ovfw0//GHeIM8YNX16W/mTJ8PQoT37eSRpgDFISJL6j7Fj4bDD8gZ5zMQDD8Ds2Z2vnj17dunxiy/CDTfkDfKzU6e2BYtp0+B1r+uZzyBJA4RBQpLUfw0ZkqeE3XPP6vcsXgwPP9x5OStWwK235g3yatyTJsH++8O55zrlrCRV4BgJSdLANnIk3HILnHdenpa2lkXuWhfZu+qqzls6JGk9ZouEJGlgGzoUZszIG+SB1n/+M9x+ex7MffvtsHBh5WdnzKi+RsUJJ+SWipkz4S1vgVGjeqDyktR3GSQkSeuXQYPaVuQ+/vi8rsWTT+ZA0brNn5/vbQ0f5V59Fb7/fVi5Ei64oK0r1IwZbcFi5Mje+kSS1BQGCUnS+i0iTy27445w9NH53PPP50BRbTrZu+7KIaJVa1eou++Gr30tj93Yay844IAcLPbd1y5SkgYcg4QkSeW22CKvkl1N66Dsal57rW2K2vPOy92rpk3LweLd74aJExtaXUlqBoOEJEn1+tSnYPz4HChuuaXrWaFWrYLbbsvbypUGCUkDgkFCkqR6bb45/PM/5w3gb3/LXaFuuQVmzYJHHqn+7AEHVD7/7LNw7bX5+u6757EcktSHGSQkSequLbeE970vb5BngWoNFbNmwYIF+fzQoXm8RCU33gif/Wz+e9NN88Dtf/qnPMZil12qzx4lSU1ikJAkqdG23ho+9KG8QZ4V6pZbcsCoNuj6llva/l68GP7rv/IGsNVWOVDMnJlbLHbYwWAhqekMEpIk9bQddshbNSnllotqnnsOfvKTvAFsu21bsJg5Mx9LUi+zA6YkSc322mvwxS/mGZ1Gj+76/qefhh/+ED7yEdhuO3jmmR6voiSVM0hIktRsG2wAJ50E11+fuzXddRecfz687W159ezO7LADvP71la898EBeE0OSeoBdmyRJ6ksGD86L2e211/9v787D5KzqRI9/fwESQkxYwzImkAQIARK2sIXIEsCAch14WNwZYa56HXcE0es4Kj4yd5Qr4K44A4gybjDiHUVFAYEBAyQhgNmQJWxBdkgCWSA594/z1nR1UdXd1f1WV6f7+3me87ypdznveXO6q+tXZ8utFOvW5cDixhtzuu22zovhzZrVOK/3vjdfO2VKHrx95JF5u+OOrX4KSUOAgYQkSQPZ8OHwhjfk9E//BGvW5IXuKoHF7Nn1r1uxAubNy/9esiSn7343v54yBQ4/PKcjjsjdoySpSQYSkiRtTDbfvGOQdVduuQXWr69/rBJYfP/7+fX48bml4ogjctp9d2eFktQtx0hIkjRYHXhgzxa2e/RR+OEP4X3vgz32gHe/u/Vlk7TRM5CQJGkwOuEEuPNOeP55+PWv4dxz4eCDexZY7Ltv/f2vvAI33QQvv1xuWSVtlOzaJEnSYDZmDLz5zTkBrFyZx1jcfHPu/nT77Z0Hb0Pu3lTPXXflLlCbbgrTp+dVug87DGbMgNe/vqWPIWngMZCQJGkoGT06D9CuDNJesybP7FQJLBYsgAMOqH/trbfm7auv5gDk9tvhoovyvvHjc0BRCS722y9Paytp0DKQkCRpKNt8845B1gAbNjTu/lQJJOp59NGcfvaz/HrkyDyF7SGH5DRrFmyzTblll9RWBhKSJKlDV2ModtkFJk2CBx/sPp/Vq3Mrx80359fXXZcX2JM0aDjYWpIk9cxXvwoPPADLl8PVV8PZZ+duTCNGdH/tQQfV3//zn8OZZ8J3vgPz5+cB3ZI2CrZISJKk5uy0E5x8ck6QB2vfdVcexH3rrXn17See6Dh/yhTYaqv6eV17LVx+eU6Qu1rtv3/uDnXQQXkK291269lsU5L6lYGEJEnqmxEj4NBDczrrLEgJHn44BxZ33AFjxza+9s47O7+urNz9p2aQOYEAABlXSURBVD917NtyyzxL1PTpObA48ECYONFF86Q2M5CQJEnlioAJE3J6xzsan7dyJSxa1H1+L74IN9yQU8W228Ijj8AWW/S1tJJ6yUBCkiS1x4gR8Ic/5FaLOXPydLJ//WvPrt1yy8ZBxFVXwfr1uQVj0iS7RUktYiAhSZLaY/hwOPronCB3iXr88RxY3H47zJsHc+fmFolajda6ADj//LweBuQF+fbbL4+7qGz32ss1LqQSGEhIkqSBIQLGjcupMpA7pTxT1Lx5eTzF3Ln5340CiTVr4M9/7ni9YkXnaWghBzB77905uNhnnxx0SOoxAwlJkjRwReRZm3bbDd72trxvw4Y8U1Q9996bV97uyrp1eZapu+7qvP/oo+H66/teZmmIsNOgJEnauAwbllfOrmfrreGcc/JK2ltu2Vy+XZ1/0UXw4x/D4sXdByrSEFFai0REjAO+CBwPbAs8AVwDnJdSer6JfLYBPgecBOwEPAv8FvhcSumxvt47Il4PnAy8GdizuMcqYD7wnZTSf9TJ/yjgxi6K/eWU0qd7+oySJKlFdtsNLrgg/zsleOih3PKwYEHH9vHH61+7//71969dC+ee2xFAbL557hq177457bNP3m69dfnPIw1gpQQSEbErcBuwPfBLYAlwMPAx4PiImJlSerYH+Wxb5DMZuAH4CTAFOBM4ISJmpJQe7OO9PwJ8CniIHBz8FdiFHFwcGxEXpZQ+0aCINwF/rLP/v7p7NkmS1M8i8qxNkybBKad07H/66c6BxV13wdKlebxEPbWtEGvW5HEa8+Z1Pm/8+I7AYtq0vJ08GTa1J7kGp7J+sr9N/iD/0ZTSNyo7I+JC4CzgfOADPcjnn8lBRKcP8xHxUeBrxX2O7+O97wCOSindVJ1JROwJzAHOiogrU0o17w4A/DGl9IUePIckSRqoxo6FN74xp4qXXmr8gb8yA1R3Hn00p1/9qmPfW98KP/1p78sqDWB9HiMREZOA2cAy4Fs1hz8PvAScHhGjuslnFHB6cf7naw5/s8j/uOJ+vb53Suk/aoOIYv9ioPKbflRXZZUkSYPMqFF5XYt6pk6FT34Sjj0Wtt++uXz33rv+/rVrYfbsvBL4ZZfl2ahefrm5vKU2K6NFopj8metSShuqD6SUVkbEreQP+4cCXU2FMAMYWeSzsiafDRFxHfB+YBZQ6d5U1r0rXim2jUZR7RYRHwbGkLtE3ZJS+ksP8pUkSRurAw/MqeLJJ+Huu+Gee/L27rsbD8LeZ5/6eS5eDL//fU4Vw4bB7rvnwKWSpk2DXXe1e5QGpDJ+Kvcotvc1OP4X8of5yXT9Yb4n+VDkU/a9iYgxwClAAq5rcNq7ilR93dXA+3o6oDwi6nWZgjwWRJIkDXQ77JBbE2bP7ti3bl0ODirBxb335n9Pm1Y/j7vvfu2+DRvyWI2lS+Hqqzv2jxgBe+6Z85o6NXfJajQwXOpHZQQSlbnS6iw72Wn/Vi3Ip5R7R0QA/wrsAHy76OZU7Wng08Cvyd2oNgcOJI/pOAXYMSKOqG0VkSRJQ8Tw4R2zOJ1+evfn1wskGlm7No/TqIzVeOWV+oHEunWwfDnsvHNu3ZBarD/ayaLYpjbk09NrvgqcBtwCvGbGppTSQmBh1a5VwG8j4jZgATATeAt51qgupZSm1y1obqlosEynJEkaVD72MZgxo6Pl4p578lS1PTF1av3999wDBx2Ux3vstVcenzF1at7uvXdeMTyi/rVSL5QRSFS+9W+0isuYmvPKzKfP946IC8izO90MnJBSarBU5mullFZExL8D/wgcQQ8CCUmSJHbZJafTTuvYt3IlLFwIf/5z5/Tkk52vbRRILCy+83zpJbjzzpyqjRmTA4xKkLH33vnfBhjqpTICiaXFdnKD47sX20bjGPqST5/uHREXAR8nryfxP1JKvZku4eli2+WsVJIkSV0aPRoOPTSnak8/nVsuFi6ERYtg4sT61y9cWH9/xYoVMGdOTtXGjIGvfQ3OOKPXRdfQVEYgUVnxeXZEDKseJxARo8ndflaT12joypzivJkRMbp65qaIGEYeNF19v17fuxgT8U3gg8DvgRNTSqt78rB1VH7bH+zyLEmSpN4YOxaOPjqn7my1FbzwQnP5r1gB22xT/9jChXDhhZ1bMWzBUKHPgURK6YFiatbZwIeAb1QdPo/8Tf33UkovVXZGxJTi2iVV+ayKiB+Sp3j9AnB2VT4fBiYAv6te2bqX9w7gEuC9wG+Ak1NKa7p6xoiYCfypdjB1RLwbeBuwDvhZV3lIkiS11Fe+Al/+MjzxRO4StXBhR1epRYty16lGGq13cccdcOmlnfeNHt0RWFRvx483wBhiyhps/UHgNuDrEXEMsBg4hLzmw33kMQTVKrMi1f60fYa8GNwnImI/8irUewInAk+Rg4W+3vtz5CBiNXmg9KfjtT/0C1JK11S9vhIYVgyufow8a9NBwMHkNSf+V0ppWZ2ySZIk9Z8I+Ju/yal6etqU4LHHOrpHVW9ffRUmTKif36JFr923ciXcfntO1SoBRiW4OOmkvAaGBq1SAomiZeBA4IvA8cCbgSeArwPnpZSe62E+z0bEDPKq1CcBhwPPApcBn0spPVbCvSsdC0cC/7tBUX4AVAcS3wGOJXeV2o4cAD0OXA5cnFJqYg43SZKkfhaRWwzGj4fjj+/YnxI89RRsskn967obd1GtNsCYMqV+IPHii3lA+E472YKxkYuU+jorq8oQEfMOOOCAA+bNa7RenSRJUj+75RaYO7ejm9SiRXlMRU889FD9lo7LL4czz8zjMiqL7FW2U6fClo0m41RZpk+fzvz58+c3Wpagp1xvXZIkSfUdfnhOFSnB44/X7yJVHWBssUVeGK+ee+/N2+eeg5tuyqnazjt3Di6mTcsrew8fXu6zqc8MJCRJktQzEXnWpnHj4LjjOvZXAoxKULFqVePVtSuBRCOPPJLTtdd27NtssxxUfO97edE9DQgGEpIkSeqb6gCjepB3PSNG5BaLl5tYvuuVV+Cuu/L0tvXcd19euG+//fKgb/ULAwlJkiT1n//8T9iwAZYty60T996bp6i9915YuhTWr69/3ejRjWeB+rd/y9PfRsDkyXDAAR1p//1h661b9jhDmYGEJEmS+tewYTBpUk4nntixf+3aHExUBxd3352nrj3ggMbdpebPz9uU8vVLl8KPf9xxfOJEmD69c4Axdmzrnm+IMJCQJEnSwDBiBOyzT07Vnn4ann++/jUp5W5PXXnooZyuuqpj37hxOaC45BLYYYe+lXuIMpCQJEnSwDZ2bOMWhNWr88Dv+fNzS0RPlzZ47DF45pnG3Z5Scp2LbhhISJIkaeO1xRZw5ZX536tW5a5Q8+fnNG9enkmq0biL6dMbTyv7vvfB/ffDUUfldOihsPnmrXiCjZaBhCRJkgaH170OZs7MqWL16jzWohJczJ+fX69bBzNm1M8nJfjNb2D58rzOxXnn5W5Xhx8OJ5yQx3VMnNg/zzSAGUhIkiRp8Bo5Eg4+OKeKdevyeheNVtG+//4cRFRbuxb+8IeczjorTzX7rnfBO94Br39968o/gDUY+i5JkiQNUsOH52lhJ02qf3zOnO7zWLAAPvlJGD8ejj0WfvCD5tbGGAQMJCRJkqRqp5+eZ3m67DJ4z3tg550bn5sSXH89nHFG7u70rW/1WzHbzUBCkiRJqjVhQg4OLr88L563dClceCEceWTj9SyeegoefLD/ythmBhKSJElSVyorZp91Fvzxj/D443DRRXnWp2ojR8K557aliO1gICFJkiQ1Y8cd4eMfh7lz86Dtc86BUaPggx8cUovbGUhIkiRJvbXXXnDBBbn702c+0/i8H/0oD9AeRJz+VZIkSeqr7bZrfKwyGPt1r4NbboFp0/qtWK1ki4QkSZLUKkuXwqmn5tW1X3wRjjsut14MAgYSkiRJUqv84hfwwgsdr594At70pkGx5oSBhCRJktQqn/oUfOQjnfctWQJf/Wp7ylMiAwlJkiSpVSLg4ovhtNM67/+Xf4Hly9tTppIYSEiSJEmtNGwYXHJJ5wHZL78Mn/1s+8pUAgMJSZIkqdW22grOO6/zviuugGeeaU95SmAgIUmSJPWH978fdt+94/X69XDtte0rTx8ZSEiSJEn9YdNNXztWYu7c9pSlBAYSkiRJUn+ZPr3z60WL2lOOEhhISJIkSf1lr706v168uD3lKIGBhCRJktRfJk7s/PrJJyGl9pSljwwkJEmSpP4yYgSMHNnxev16WLWqfeXpg03bXQBJkiRpSJk4EZ57DjbZJKfVq2H06HaXqmkGEpIkSVJ/Wriw3SUohV2bJEmSJDXNQEKSJElS0wwkJEmSJDXNQEKSJElS0wwkJEmSJDXNWZskSZKk/nT//bBmTcfrXXftvLbERsJAQpIkSepPu+3W7hKUwq5NkiRJkppWWiAREeMi4tKIWB4RayNiWURcHBFbN5nPNsV1y4p8lhf5jivz3hGxV0T8LCKeiog1EbE0Is6LiIbtShFxWERcGxHPRcTLEXFPRHw8IjZp5hklSZKkjV0pXZsiYlfgNmB74JfAEuBg4GPA8RExM6X0bA/y2bbIZzJwA/ATYApwJnBCRMxIKT3Y13tHxCFF/psBVwGPAkcDnwOOiYhjUkpra645EbgaWAP8FHgOeAtwETATOK37/ylJkiRpcCirReLb5A/yH00pnZRS+nRK6Wjyh+w9gPN7mM8/k4OIi1JKxxT5nEQOCrYv7tOnexetB5cBWwCnppTemVL6FHAIOVCYCZxVc80Y4PvAeuColNL/TCl9EtgP+BNwakS8vYfPKEmSJG30+hxIRMQkYDawDPhWzeHPAy8Bp0fEqG7yGQWcXpz/+ZrD3yzyP664X1/ufSSwJ3BzSun/VXamlDYA5xYvPxARUXXNqcBY4CcppblV16wBPlu8/Ieunk+SJEkaTMpokTi62F5XfBj/bymllcCt5G//D+0mnxnASODW4rrqfDYA1xUvZ/Xx3pVrfltbgKLb1H3ALsCknlwD3Ay8DBwWESPqPpkkSZI0yJQRSOxRbO9rcPwvxXZyC/Jp+zUppVeBh8jjTSbVHpckSZIGozIGW29ZbF9scLyyf6sW5DOQr6krIuY1OLTv4sWLmT59endZSJIkSb22ePFigAl9zac/FqSrjDVIbchnIF9Ta/3q1atfnD9//rI+5NEbU4rtkn6+r/qX9Tw0WM9Dg/U8+FnHQ0M763kCsKKvmZQRSFS+jd+ywfExNeeVmc9AvqaulNKAanKotJAMtHKpXNbz0GA9Dw3W8+BnHQ8Ng6GeyxgjsbTYNhoDsXuxbTQmoS/5tP2aiNgUmAi8CjxYe1ySJEkajMoIJG4strMjolN+ETGavC7DamBON/nMKc6bWVxXnc8w8jSv1ffr7b1vKLbH1xagmE52MvAwnYOChtcAR5BnhrqtdhE7SZIkabDqcyCRUnqAPDXrBOBDNYfPA0YBV6SUXqrsjIgpETGl+sSU0irgh8X5X6jJ58NF/r+rXtm6N/cGbgIWA0dExN9WlWkY8OXi5XdTStXjHa4CngHeHhEHVl2zOfCl4uV3kCRJkoaI6Px5uZeZROwK3EZeYfqX5A/qh5DXfLgPOCyl9GzV+QkgpRQ1+Wxb5DOZ3ApwB3nxuBOBp4p8HujLvYtrDiny34wcJDwCHAMcSF574pja1oWIOKk4dw3wE+A54G/JU8NeBbw1lfGf2c8GQ/88dc96Hhqs56HBeh78rOOhYTDUcxldmyotAwcCl5M/xJ8N7Ap8HZhR+0G+i3yeJS9M93VgtyKfQ4DLgOm1QURv751Suh04iBx4zAbOIg+k/iLwxnpdlFJK15BXxb4ZOAX4CPAK8Ang7RtjECFJkiT1ViktEpIkSZKGllJaJCRJkiQNLQYSkiRJkppmICFJkiSpaQYSkiRJkppmICFJkiSpaQYSkiRJkppmICFJkiSpaQYSg1BEjIuISyNieUSsjYhlEXFxRGzdZD7bFNctK/JZXuQ7rlVlV8/1tZ4jYlREvCsi/j0ilkTESxGxMiLmRsTZETG81c+grpX1u1yT5xERsT4iUkR8qczyqnfKrOeImBYRV0TEo0VeT0XETRHxd60ou3quxL/Nb4iIXxbXr4mIRyLi2og4vlVlV/ci4tSI+EZE3BIRK4r32B/1Mq/S3/tbxQXpBpmI2BW4DdievHL3EuBgYBawFJjZk5XGI2LbIp/JwA3AncAU4ETgKfKq4Q+24hnUvTLqufij8xvgOeBG4H5gG+AtwI5F/seklNa06DHUhbJ+l2vyHA3cA2wHvA44P6X02TLLreaUWc8RcQbwr8DLwK+AZcBWwFRgeUrp7SUXXz1U4t/mfwC+DbwE/AJ4DBgHnAxsAXw2pXR+K55BXYuIBcC+wCpyvUwBrkwpvbvJfEp/72+plJJpECXgd0ACPlKz/8Ji/3d7mM/3ivMvrNn/0WL/b9v9rEM5lVHPwH7Au4DhNftHA/OKfM5u97MO1VTW73LNtZeSA8fPFHl8qd3POdRTie/ZhwKvAguAHesc36zdzzqUU0nv2ZsBLwCrgT1qju0JrCEHkSPa/bxDMZE/6O8OBHBUUa8/asfPSn8mWyQGkYiYBDxA/hZq15TShqpjo4EnyD/g26eUXuoin1HA08AGYKeU0sqqY8OKe0wo7mGrRD8rq567ucc7gSuBX6WU3tLnQqsprajjiDgRuAY4HdgUuAxbJNqqzHqOiJuBw4FpKaU/t6zQalqJf5t3AP4K3JNS2rfO8XuAacB2aSB9Yz0ERcRR5Jb+plok+uPve9kcIzG4HF1sr6v+4QMogoFbyU2fh3aTzwxgJHBrdRBR5LMBuK54OavPJVZvlFXPXXml2L7ahzzUe6XWcURsD3wfuCal1Ks+u2qJUuq5GLd2ODAXWBgRsyLinGKs0zHFF0Bqn7J+n58if8k3OSJ2rz4QEZPJ34YvMIjYqPXH3/dS+eYyuOxRbO9rcPwvxXZyP+Wj1uiP+vn7YvvbPuSh3iu7ji8hv99/oC+FUunKqueDqs6/oUgXAP8X+AOwICJ260M51Tel1HPKXUg+RP5dnhcRP4iI/xMRV5C7oy4ETiuhvGqfje7z16btLoBKtWWxfbHB8cr+rfopH7VGS+snIj4MHE/ua31pb/JQn5VWxxHx9+RJEt6WUnqyhLKpPGXV8/bF9q3AM+SBt9cDY4HPk7uz/ToipqWU1vW+uOql0n6fU0o/j4jlwI+B6pm4niR3V7S78cZto/v8ZYvE0BLFtq8DY8rKR63R6/qJiJOBi8n9cE9JKb3SzSVqjx7VcURMINfnz1NKP2txmVS+nv4ub1K1fW9K6RcppRUppQeA95C7PE0GTmlNMdVHPX7Pjoh3k1uZbiEPsN6i2F4PfBP4SYvKqIFhwH3+MpAYXCqR6pYNjo+pOa/V+ag1WlI/EXES+Y/QU8BRDqRvq7Lq+FLyDC8fLKNQKl1Z9fx8sV0LXFt9oOgO88vi5cHNFlClKKWei3EQl5K7MJ2eUlqSUlqdUlpCbnWaB5xWDPTVxmmj+/xlIDG4LC22jfrOVQZnNep7V3Y+ao3S6yciTgN+Tm4ePzKltLSbS9RaZdXxAeRuL08XiyOliEjkLhAA/1jsu6ZvxVUvlf2evbJ2gGahEmiMbKJsKk9Z9TybPAXsTXUG4m4Abi5eTu9NITUgbHSfvxwjMbjcWGxnR8SwOtOGzSR/Ozmnm3zmFOfNjIjRdaZ/nV1zP/Wvsuq5cs07gSuAx4FZtkQMCGXV8RXkrg+1dgeOII+DmQfc1ecSqzfKqud7yGMjtouIHeqMhZlabJf1vcjqhbLqeUSxHdvgeGW/42A2XqX+fe8PtkgMIkV/2OvIazx8qObwecAo4IrquYcjYkpETKnJZxXww+L8L9Tk8+Ei/9/5gbM9yqrnYv97yHX9CHCEdTowlPi7/NGU0ntrEx0tEr8u9n2rZQ+jhkqs51fJi4gCfKV6uteImAacQZ7K+aqSH0E9UOJ79i3F9tSI2Kf6QETsB5xK7jt/Q3mlVytExGZFHe9avb83Pyvt5oJ0g0ydpdUXA4eQ13y4Dziseo7popsDKaWoyWfbIp/J5DelO8gDuk4k96E/rPiBVxuUUc8RMYs8aG8Yud/to3Vu9UJK6eIWPYa6UNbvcoO8z8AF6QaEEt+ztyAPuD2U3ML0R/I31KeQuzSdnVK6sMWPowZKrOdLgTPJrQ6/AB4mf+g8CRgOXJxSOqvFj6M6inGGJxUvdwSOI8+iVQkAn0kpnVOcOwF4CHg4pTShJp+mflbarqwlsk0DJwHjyR8SniC/2TwMfA3Yps65iWI8Xp1j2xTXPVzk8wT5A+e4dj+jqe/1TP6WMnWTlrX7OYdyKut3uc65lbr/Uruf0VTqe/YW5FbkJeSB1y+Svyx4U7uf0VROPZNn7TmDHCg+T25peo4cRL693c84lFPxu9ejv6fk4K/h39hmflbanWyRkCRJktQ0x0hIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJapqBhCRJkqSmGUhIkiRJatr/BzksvxAEZptzAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<matplotlib.figure.Figure at 0x21a9198d550>"
            ]
          },
          "metadata": {
            "tags": [],
            "image/png": {
              "height": 252,
              "width": 393
            }
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UcdGWoPiyC8K",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}