deep_learning_keras.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "deep-learning fastai",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 106
        },
        "id": "ECypRwBsU0F1",
        "outputId": "1cf1f177-def9-4eb0-f0d6-635b7d5ec32a"
      },
      "source": [
        "# Upload data from local drive\n",
        "from google.colab import files\n",
        "uploaded = files.upload()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-f1a4586e-5a13-4c4d-9512-e49aa3add523\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-f1a4586e-5a13-4c4d-9512-e49aa3add523\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving test.csv to test.csv\n",
            "Saving train4.csv to train4.csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0QnZczol5Xs3"
      },
      "source": [
        "## Keras （User-Product pairs)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UcwPJN1855SM"
      },
      "source": [
        "import keras\n",
        "from keras.models import Sequential, load_model\n",
        "from keras.layers import Dense, Dropout, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D, BatchNormalization\n",
        "from keras.optimizers import Adam,SGD,Adagrad,Adadelta,RMSprop\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.callbacks import ReduceLROnPlateau, LearningRateScheduler\n",
        "from keras.utils import to_categorical"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "m_BWAUhh5__c"
      },
      "source": [
        "from keras.layers import Embedding, Input, dot, concatenate\n",
        "from keras.models import Model\n",
        "from IPython.display import SVG\n",
        "#from keras.utils.vis_utils import model_to_do"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "88uU704S7nKO"
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "df = pd.read_csv('train4.csv') # train\n",
        "#df_val = pd.read_csv('validation.csv') # validation"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KZYphlBn78py",
        "outputId": "9ea6b508-d930-43a5-cf68-088068a47b42"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "\n",
        "user_enc = LabelEncoder()\n",
        "df['user_id'] = user_enc.fit_transform(df['user_id'].values)\n",
        "n_users = df['user_id'].nunique()\n",
        "item_enc = LabelEncoder()\n",
        "df['product_id'] = item_enc.fit_transform(df['product_id'].values)\n",
        "n_products = df['product_id'].nunique()\n",
        "df['purchase_count'] = df['purchase_count'].values.astype(np.float32)\n",
        "min_cnt = min(df['purchase_count'])\n",
        "max_cnt = max(df['purchase_count'])\n",
        "n_users, n_products , min_cnt, max_cnt\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(37284, 36847, 1.0, 18.0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "CmhP7eH36C3t"
      },
      "source": [
        "X = df[['user_id', 'product_id']].values\n",
        "Y = df['purchase_count'].values\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "x_train1, x_test, y_train1, y_test = train_test_split(X, Y, test_size = 0.2, random_state = 42)\n",
        "x_train, x_val, y_train, y_val = train_test_split(x_train1, y_train1, test_size = 0.2, random_state = 42)\n"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PPTwtdU_9kzM"
      },
      "source": [
        "#x_train.shape, x_val.shape, y_train.shape, y_val.shape"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x0223BEe6K_f"
      },
      "source": [
        "# embeddings\n",
        "\n",
        "num_latent_factors = 30\n",
        "\n",
        "X_train_array = [x_train[:, 0], x_train[:, 1]]\n",
        "X_test_array = [x_test[:, 0], x_test[:, 1]]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8PlST_1z95LT"
      },
      "source": [
        "from keras.models import Model\n",
        "from keras.layers import Input, Reshape, Dot\n",
        "from keras.layers.embeddings import Embedding\n",
        "from keras.optimizers import Adam\n",
        "from keras.regularizers import l2\n",
        "def RecommenderV1(n_users, n_products, n_factors):\n",
        "    user = Input(shape=(1,))\n",
        "    u = Embedding(n_users, n_factors, embeddings_initializer='he_normal',\n",
        "                  embeddings_regularizer=l2(1e-6))(user)\n",
        "    u = Reshape((n_factors,))(u)\n",
        "    \n",
        "    product = Input(shape=(1,))\n",
        "    m = Embedding(n_products, n_factors, embeddings_initializer='he_normal',\n",
        "                  embeddings_regularizer=l2(1e-6))(product)\n",
        "    m = Reshape((n_factors,))(m)\n",
        "    \n",
        "    x = Dot(axes=1)([u, m])\n",
        "    model = Model(inputs=[user, product], outputs=x)\n",
        "    opt = Adam(lr=0.001)\n",
        "    model.compile(loss='mean_squared_error', optimizer=opt, metrics = ['acc'])\n",
        "    return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hJq_YiSE-SK6"
      },
      "source": [
        "\n",
        "n_factors = 30"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Bdxhybsk95Hk",
        "outputId": "219b3b20-4b19-421d-8a4f-0d67764cc707"
      },
      "source": [
        "model = RecommenderV1(n_users, n_products, n_factors)\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"functional_1\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "input_2 (InputLayer)            [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "embedding (Embedding)           (None, 1, 30)        1118520     input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "embedding_1 (Embedding)         (None, 1, 30)        1105410     input_2[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "reshape (Reshape)               (None, 30)           0           embedding[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "reshape_1 (Reshape)             (None, 30)           0           embedding_1[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "dot (Dot)                       (None, 1)            0           reshape[0][0]                    \n",
            "                                                                 reshape_1[0][0]                  \n",
            "==================================================================================================\n",
            "Total params: 2,223,930\n",
            "Trainable params: 2,223,930\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "584LAiBO95Dv",
        "outputId": "c86f78d2-e4d9-4b48-844e-a8ef54d1544f"
      },
      "source": [
        "history = model.fit(x=X_train_array, y=y_train, batch_size=2048, epochs=30, verbose=1,\n",
        "                    validation_data=(X_test_array, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/30\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/indexed_slices.py:432: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
            "  \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "232/232 [==============================] - 4s 15ms/step - loss: 1.1999 - acc: 0.0000e+00 - val_loss: 1.1692 - val_acc: 0.0000e+00\n",
            "Epoch 2/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.8999 - acc: 0.0808 - val_loss: 0.5808 - val_acc: 0.3120\n",
            "Epoch 3/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.3948 - acc: 0.5684 - val_loss: 0.3039 - val_acc: 0.7056\n",
            "Epoch 4/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.2313 - acc: 0.8029 - val_loss: 0.2207 - val_acc: 0.8288\n",
            "Epoch 5/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1764 - acc: 0.8787 - val_loss: 0.1882 - val_acc: 0.8763\n",
            "Epoch 6/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1528 - acc: 0.9098 - val_loss: 0.1733 - val_acc: 0.8983\n",
            "Epoch 7/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1408 - acc: 0.9245 - val_loss: 0.1658 - val_acc: 0.9102\n",
            "Epoch 8/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1339 - acc: 0.9332 - val_loss: 0.1617 - val_acc: 0.9173\n",
            "Epoch 9/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1294 - acc: 0.9376 - val_loss: 0.1595 - val_acc: 0.9206\n",
            "Epoch 10/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1261 - acc: 0.9401 - val_loss: 0.1582 - val_acc: 0.9227\n",
            "Epoch 11/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1236 - acc: 0.9423 - val_loss: 0.1576 - val_acc: 0.9250\n",
            "Epoch 12/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1214 - acc: 0.9444 - val_loss: 0.1574 - val_acc: 0.9273\n",
            "Epoch 13/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1194 - acc: 0.9465 - val_loss: 0.1574 - val_acc: 0.9297\n",
            "Epoch 14/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1176 - acc: 0.9483 - val_loss: 0.1577 - val_acc: 0.9316\n",
            "Epoch 15/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1159 - acc: 0.9496 - val_loss: 0.1580 - val_acc: 0.9328\n",
            "Epoch 16/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1142 - acc: 0.9503 - val_loss: 0.1585 - val_acc: 0.9335\n",
            "Epoch 17/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1126 - acc: 0.9507 - val_loss: 0.1589 - val_acc: 0.9341\n",
            "Epoch 18/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1110 - acc: 0.9510 - val_loss: 0.1597 - val_acc: 0.9342\n",
            "Epoch 19/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1094 - acc: 0.9511 - val_loss: 0.1604 - val_acc: 0.9344\n",
            "Epoch 20/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1079 - acc: 0.9512 - val_loss: 0.1610 - val_acc: 0.9343\n",
            "Epoch 21/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1064 - acc: 0.9512 - val_loss: 0.1617 - val_acc: 0.9343\n",
            "Epoch 22/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1049 - acc: 0.9512 - val_loss: 0.1626 - val_acc: 0.9342\n",
            "Epoch 23/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1035 - acc: 0.9513 - val_loss: 0.1634 - val_acc: 0.9342\n",
            "Epoch 24/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1021 - acc: 0.9513 - val_loss: 0.1642 - val_acc: 0.9340\n",
            "Epoch 25/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.1007 - acc: 0.9513 - val_loss: 0.1650 - val_acc: 0.9338\n",
            "Epoch 26/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.0994 - acc: 0.9513 - val_loss: 0.1659 - val_acc: 0.9337\n",
            "Epoch 27/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.0981 - acc: 0.9513 - val_loss: 0.1668 - val_acc: 0.9335\n",
            "Epoch 28/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.0969 - acc: 0.9513 - val_loss: 0.1676 - val_acc: 0.9333\n",
            "Epoch 29/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.0957 - acc: 0.9513 - val_loss: 0.1685 - val_acc: 0.9331\n",
            "Epoch 30/30\n",
            "232/232 [==============================] - 3s 15ms/step - loss: 0.0945 - acc: 0.9513 - val_loss: 0.1694 - val_acc: 0.9331\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "qaYSPliW946V",
        "outputId": "9eaf19a9-e863-471d-a4d7-ffc5ffd23bda"
      },
      "source": [
        "%matplotlib inline\n",
        "training_loss = history.history['loss']\n",
        "test_loss = history.history['val_loss']\n",
        "\n",
        "# Create count of the number of epochs\n",
        "epoch_count = range(1, len(training_loss) + 1)\n",
        "\n",
        "# Visualize loss history\n",
        "plt.figure(figsize = (8,4))\n",
        "plt.plot(epoch_count, training_loss, 'r--')\n",
        "plt.plot(epoch_count, test_loss, 'b-')\n",
        "plt.legend(['Training Loss', 'Val Loss'])\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LT161ViCCmMo",
        "outputId": "85692dba-9011-4e82-f01d-1926c3b9643a"
      },
      "source": [
        "model.evaluate(X_test_array, y_test,verbose=0)\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.16937674582004547, 0.9330628514289856]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 21
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "88y8xDs2sR8x"
      },
      "source": [
        "y_pred1 = model.predict(X_test_array)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XM6x3cPZsRyz"
      },
      "source": [
        "def precisionatk(y_true,y_pred, k):\n",
        "    precision_average = []\n",
        "    idx =  (-y_pred).argsort(axis=-1)[:,:k] ###k\n",
        "    for i in range(idx.shape[0]):\n",
        "        precision_sample = 0\n",
        "        for j in idx[i,:]:\n",
        "            if y_true[j] > 0:\n",
        "                precision_sample += 1\n",
        "        precision_sample = precision_sample / k\n",
        "        precision_average.append(precision_sample)\n",
        "    return np.mean(precision_average)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "y1IdgpU_tks8"
      },
      "source": [
        "pre_k2 = []\n",
        "for k in range(2, 31):\n",
        "    pre = precisionatk(y_test, y_pred1, k)\n",
        "    pre_k2.append(pre)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cKwrGv659zVw",
        "outputId": "7843688c-e1dc-4d12-c89e-03b798825c3d"
      },
      "source": [
        "np.mean(pre_k2)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "0.10327541830759966"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 32
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "t6NV34E5tvZd",
        "outputId": "2d6bb4cb-bf3d-40c9-950f-31f45fa16530"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(range(2, 31), pre_k2)\n",
        "plt.xlabel('k')\n",
        "plt.ylabel('precision@K')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7CAPV0WSP3Ow"
      },
      "source": [
        "# User- brand pair (this)"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "resources": {
            "http://localhost:8080/nbextensions/google.colab/files.js": {
              "data": "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",
              "ok": true,
              "headers": [
                [
                  "content-type",
                  "application/javascript"
                ]
              ],
              "status": 200,
              "status_text": ""
            }
          },
          "base_uri": "https://localhost:8080/",
          "height": 72
        },
        "id": "TjWd3nY8PXhx",
        "outputId": "e2df66b3-6980-4439-c3f5-38fbcccf9316"
      },
      "source": [
        "from google.colab import files\n",
        "uploaded = files.upload()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "\n",
              "     <input type=\"file\" id=\"files-c1a3e370-26a8-417b-8eb8-fa6ddfcfec67\" name=\"files[]\" multiple disabled\n",
              "        style=\"border:none\" />\n",
              "     <output id=\"result-c1a3e370-26a8-417b-8eb8-fa6ddfcfec67\">\n",
              "      Upload widget is only available when the cell has been executed in the\n",
              "      current browser session. Please rerun this cell to enable.\n",
              "      </output>\n",
              "      <script src=\"/nbextensions/google.colab/files.js\"></script> "
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "stream",
          "text": [
            "Saving user-brand-cnt-3-months.csv to user-brand-cnt-3-months.csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LNZ6c7NVi_2U",
        "outputId": "1c18fc58-7280-4d2e-9809-871ffc072d76"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-NDGb9cgP1Y-"
      },
      "source": [
        "import pandas as pd\n",
        "\n",
        "df2 = pd.read_csv('user-brand-cnt-3-months.csv')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QKZ7byc8PjYH"
      },
      "source": [
        "count = df2['purchase_count']\n",
        "df2 =df2.drop(['purchase_count', 'brand'], axis = 1)\n",
        "df2['purchase_count'] = count"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 419
        },
        "id": "ymLxC94QQBlr",
        "outputId": "2bb516c6-9ad0-4a06-b16e-572f98c6b6ad"
      },
      "source": [
        "df2"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>user_id</th>\n",
              "      <th>brand_id</th>\n",
              "      <th>purchase_count</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>10280338</td>\n",
              "      <td>0</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>28129653</td>\n",
              "      <td>0</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>72062431</td>\n",
              "      <td>0</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>105697630</td>\n",
              "      <td>0</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>126051562</td>\n",
              "      <td>0</td>\n",
              "      <td>0.666667</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>205879</th>\n",
              "      <td>594355111</td>\n",
              "      <td>249</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>205880</th>\n",
              "      <td>594391949</td>\n",
              "      <td>249</td>\n",
              "      <td>0.666667</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>205881</th>\n",
              "      <td>594426448</td>\n",
              "      <td>249</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>205882</th>\n",
              "      <td>595026228</td>\n",
              "      <td>249</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>205883</th>\n",
              "      <td>595040378</td>\n",
              "      <td>249</td>\n",
              "      <td>0.333333</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>205884 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "          user_id  brand_id  purchase_count\n",
              "0        10280338         0        0.333333\n",
              "1        28129653         0        0.333333\n",
              "2        72062431         0        0.333333\n",
              "3       105697630         0        0.333333\n",
              "4       126051562         0        0.666667\n",
              "...           ...       ...             ...\n",
              "205879  594355111       249        0.333333\n",
              "205880  594391949       249        0.666667\n",
              "205881  594426448       249        0.333333\n",
              "205882  595026228       249        0.333333\n",
              "205883  595040378       249        0.333333\n",
              "\n",
              "[205884 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 269
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "grJzRa-iQTHJ",
        "outputId": "9cc5d5cf-9f5f-413e-a212-6d85273d73da"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "\n",
        "user_enc = LabelEncoder()\n",
        "df2['user_id'] = user_enc.fit_transform(df2['user_id'].values)\n",
        "n_users = df2['user_id'].nunique()\n",
        "item_enc = LabelEncoder()\n",
        "df2['brand_id'] = item_enc.fit_transform(df2['brand_id'].values)\n",
        "n_brands = df2['brand_id'].nunique()\n",
        "df2['cnt'] = df2['purchase_count'].values.astype(np.float32)\n",
        "min_cnt = min(df2['purchase_count'])\n",
        "max_cnt = max(df2['purchase_count'])\n",
        "n_users, n_brands , min_cnt, max_cnt\n"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(64262, 234, 0.3333333333333333, 53.0)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 270
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6Ls8xOSxQrpQ"
      },
      "source": [
        "X = df2[['user_id', 'brand_id']].values\n",
        "Y = df2['purchase_count'].values\n",
        "\n",
        "from sklearn.model_selection import train_test_split\n",
        "x_train, x_test, y_train, y_test = train_test_split(X, Y, test_size = 0.2, random_state = 42) "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "tBvBsgfWQrpR",
        "outputId": "e781ad5e-06a4-4c7c-d2fb-10d02d697cce"
      },
      "source": [
        "x_train.shape, x_test.shape, y_train.shape, y_test.shape"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "((164707, 2), (41177, 2), (164707,), (41177,))"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 272
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gyU5fdfcQrpR"
      },
      "source": [
        "# embeddings\n",
        "\n",
        "num_latent_factors = 30\n",
        "\n",
        "X_train_array = [x_train[:, 0], x_train[:, 1]]\n",
        "X_test_array = [x_test[:, 0], x_test[:, 1]]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FizI0n-jQrpR"
      },
      "source": [
        "from keras.models import Model\n",
        "from keras.layers import Input, Reshape, Dot\n",
        "from keras.layers.embeddings import Embedding\n",
        "from keras.optimizers import Adam\n",
        "from keras.regularizers import l2\n",
        "def RecommenderV1(n_users, n_products, n_factors):\n",
        "    user = Input(shape=(1,))\n",
        "    u = Embedding(n_users, n_factors, embeddings_initializer='he_normal',\n",
        "                  embeddings_regularizer=l2(1e-6))(user)\n",
        "    u = Reshape((n_factors,))(u)\n",
        "    \n",
        "    product = Input(shape=(1,))\n",
        "    m = Embedding(n_products, n_factors, embeddings_initializer='he_normal',\n",
        "                  embeddings_regularizer=l2(1e-6))(product)\n",
        "    m = Reshape((n_factors,))(m)\n",
        "    \n",
        "    x = Dot(axes=1)([u, m])\n",
        "    model = Model(inputs=[user, product], outputs=x)\n",
        "    opt = Adam(lr=0.001)\n",
        "    model.compile(loss='mean_squared_error', optimizer=opt, metrics = ['acc',mean_pred])\n",
        "    return model"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "13YANa51QrpR"
      },
      "source": [
        "\n",
        "n_factors = 30"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ljnQKghDQrpR",
        "outputId": "bada1652-84dc-4f86-85bf-995354326e08"
      },
      "source": [
        "model = RecommenderV1(n_users, n_brands, n_factors)\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"functional_33\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_33 (InputLayer)           [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "input_34 (InputLayer)           [(None, 1)]          0                                            \n",
            "__________________________________________________________________________________________________\n",
            "embedding_32 (Embedding)        (None, 1, 30)        1927860     input_33[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "embedding_33 (Embedding)        (None, 1, 30)        7020        input_34[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "reshape_32 (Reshape)            (None, 30)           0           embedding_32[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "reshape_33 (Reshape)            (None, 30)           0           embedding_33[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "dot_16 (Dot)                    (None, 1)            0           reshape_32[0][0]                 \n",
            "                                                                 reshape_33[0][0]                 \n",
            "==================================================================================================\n",
            "Total params: 1,934,880\n",
            "Trainable params: 1,934,880\n",
            "Non-trainable params: 0\n",
            "__________________________________________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RvrgvtmJQrpR",
        "outputId": "3229cbb0-1450-4090-9d57-95aa25bdd99f"
      },
      "source": [
        "history = model.fit(x=X_train_array, y=y_train, batch_size=16384, epochs=20, verbose=1,\n",
        "                    validation_data=(X_test_array, y_test))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/indexed_slices.py:432: UserWarning: Converting sparse IndexedSlices to a dense Tensor of unknown shape. This may consume a large amount of memory.\n",
            "  \"Converting sparse IndexedSlices to a dense Tensor of unknown shape. \"\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "11/11 [==============================] - 0s 29ms/step - loss: 1.4766 - acc: 0.0000e+00 - mean_pred: 1.9017e-05 - val_loss: 1.4835 - val_acc: 0.0000e+00 - val_mean_pred: 9.8222e-05\n",
            "Epoch 2/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.4694 - acc: 0.0000e+00 - mean_pred: 0.0040 - val_loss: 1.4827 - val_acc: 0.0000e+00 - val_mean_pred: 5.4223e-04\n",
            "Epoch 3/20\n",
            "11/11 [==============================] - 0s 17ms/step - loss: 1.4609 - acc: 0.0000e+00 - mean_pred: 0.0086 - val_loss: 1.4812 - val_acc: 0.0000e+00 - val_mean_pred: 0.0015\n",
            "Epoch 4/20\n",
            "11/11 [==============================] - 0s 17ms/step - loss: 1.4499 - acc: 0.0000e+00 - mean_pred: 0.0148 - val_loss: 1.4786 - val_acc: 0.0000e+00 - val_mean_pred: 0.0031\n",
            "Epoch 5/20\n",
            "11/11 [==============================] - 0s 17ms/step - loss: 1.4356 - acc: 0.0000e+00 - mean_pred: 0.0229 - val_loss: 1.4745 - val_acc: 0.0000e+00 - val_mean_pred: 0.0058\n",
            "Epoch 6/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.4175 - acc: 0.0000e+00 - mean_pred: 0.0334 - val_loss: 1.4686 - val_acc: 0.0000e+00 - val_mean_pred: 0.0096\n",
            "Epoch 7/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.3954 - acc: 0.0000e+00 - mean_pred: 0.0467 - val_loss: 1.4607 - val_acc: 0.0000e+00 - val_mean_pred: 0.0148\n",
            "Epoch 8/20\n",
            "11/11 [==============================] - 0s 17ms/step - loss: 1.3690 - acc: 0.0000e+00 - mean_pred: 0.0628 - val_loss: 1.4505 - val_acc: 0.0000e+00 - val_mean_pred: 0.0217\n",
            "Epoch 9/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.3387 - acc: 0.0000e+00 - mean_pred: 0.0818 - val_loss: 1.4380 - val_acc: 0.0000e+00 - val_mean_pred: 0.0302\n",
            "Epoch 10/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.3049 - acc: 0.0000e+00 - mean_pred: 0.1040 - val_loss: 1.4232 - val_acc: 0.0000e+00 - val_mean_pred: 0.0406\n",
            "Epoch 11/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.2677 - acc: 0.0000e+00 - mean_pred: 0.1291 - val_loss: 1.4064 - val_acc: 0.0000e+00 - val_mean_pred: 0.0527\n",
            "Epoch 12/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.2279 - acc: 0.0000e+00 - mean_pred: 0.1571 - val_loss: 1.3878 - val_acc: 0.0000e+00 - val_mean_pred: 0.0665\n",
            "Epoch 13/20\n",
            "11/11 [==============================] - 0s 17ms/step - loss: 1.1863 - acc: 0.0000e+00 - mean_pred: 0.1874 - val_loss: 1.3679 - val_acc: 0.0000e+00 - val_mean_pred: 0.0818\n",
            "Epoch 14/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.1437 - acc: 0.0000e+00 - mean_pred: 0.2195 - val_loss: 1.3472 - val_acc: 0.0000e+00 - val_mean_pred: 0.0983\n",
            "Epoch 15/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.1007 - acc: 1.2750e-04 - mean_pred: 0.2530 - val_loss: 1.3261 - val_acc: 0.0000e+00 - val_mean_pred: 0.1157\n",
            "Epoch 16/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.0584 - acc: 0.0032 - mean_pred: 0.2867 - val_loss: 1.3055 - val_acc: 4.8571e-05 - val_mean_pred: 0.1333\n",
            "Epoch 17/20\n",
            "11/11 [==============================] - 0s 18ms/step - loss: 1.0172 - acc: 0.0185 - mean_pred: 0.3210 - val_loss: 1.2855 - val_acc: 7.0428e-04 - val_mean_pred: 0.1509\n",
            "Epoch 18/20\n",
            "11/11 [==============================] - 0s 19ms/step - loss: 0.9775 - acc: 0.0363 - mean_pred: 0.3549 - val_loss: 1.2668 - val_acc: 0.0019 - val_mean_pred: 0.1680\n",
            "Epoch 19/20\n",
            "11/11 [==============================] - 0s 19ms/step - loss: 0.9395 - acc: 0.0496 - mean_pred: 0.3865 - val_loss: 1.2497 - val_acc: 0.0037 - val_mean_pred: 0.1842\n",
            "Epoch 20/20\n",
            "11/11 [==============================] - 0s 19ms/step - loss: 0.9035 - acc: 0.0594 - mean_pred: 0.4168 - val_loss: 1.2341 - val_acc: 0.0061 - val_mean_pred: 0.1991\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "-Pj_cMPyQrpR",
        "outputId": "c7f13374-a7a3-4bbc-91ad-6f9cd5d29646"
      },
      "source": [
        "%matplotlib inline\n",
        "training_loss = history.history['loss']\n",
        "test_loss = history.history['val_loss']\n",
        "\n",
        "# Create count of the number of epochs\n",
        "epoch_count = range(1, len(training_loss) + 1)\n",
        "\n",
        "# Visualize loss history\n",
        "plt.figure(figsize = (8,4))\n",
        "plt.plot(epoch_count, training_loss, 'r--')\n",
        "plt.plot(epoch_count, test_loss, 'b-')\n",
        "plt.legend(['Training Loss', 'Test Loss'])\n",
        "plt.xlabel('Epoch')\n",
        "plt.ylabel('Loss')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfEAAAEGCAYAAAB1pazcAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3deXiU5dXH8e8hgYRFAdkUEMEFFBABIyBLgeKCgOKuiG9di6gVcataW0VfN94qVkSlaAGtdUVABXcUQXEBRBQUxSIqqAjIFpAt3O8fZ2JiTEICmXlmkt/nup4rM888k5zJJDm5t3NbCAERERFJPZWiDkBERER2jZK4iIhIilISFxERSVFK4iIiIilKSVxERCRFpUcdQGnVrVs3NG3aNOowREREEmbu3LmrQgj1Cp6PWxI3s7FAP+DHEELrQh7vATwHfBU7NTGEcMvOPm/Tpk2ZM2dOWYYqIiKS1Mzs68LOx7MlPh4YBTxazDUzQwj94hiDiIhIuRW3MfEQwgzgp3h9fhERkYou6oltR5rZfDN7ycxaRRyLiIhISolyYtuHwH4hhGwz6wNMBg4q7EIzGwQMAmjSpEniIhQRKee2bdvGsmXL2Lx5c9ShCJCZmUnjxo2pXLlyia6PLImHENbnu/2imT1gZnVDCKsKuXYMMAYgKytLxd5FRMrIsmXL2GOPPWjatClmFnU4FVoIgdWrV7Ns2TKaNWtWoudE1p1uZntb7CfGzDrEYlkdVTwiIhXR5s2bqVOnjhJ4EjAz6tSpU6pekXguMXsC6AHUNbNlwE1AZYAQwmjgVOBiM9sO/AycGbSlmohIwimBJ4/SvhdxS+IhhAE7eXwUvgQtMkuWwLhxkJbmR3p63u2SHkU9p3JlyMiAKlV+fRQ8l54O+v0REZFdkXIV28rSV1/BbbdBlO1/s50n+tz7GRlQvbof1arl3S7N/WrV/J8MEZFksHr1anr16gXADz/8QFpaGvXqeWGyDz74gCpVqhT53Dlz5vDoo48ycuTIYr9G586dmTVr1m7HOn36dO666y6mTJmy25+rrFToJN6rF+xYs45QKY2cqjXYvh1yckp3FPWcbdtg69a8Y8uW0t0v7NzatfDdd7BxI2za5B83biz9PyGZmXlJvUYNqFkTateGWrX8Y+6R/37+23vsAZWiXpwoIuVCnTp1+OijjwAYNmwYNWrU4Oqrr/7l8e3bt5OeXniqysrKIisra6dfoywSeLKq0EkcgHvuwW6+mfTatUnfd19o0sSPu+/2bPfNN54lGzb0PvIkEwJs3vzrpF4wyRf3WHY2rFsHK1fCF1/4Pwpr1/o/IkWpVCkv8ReV7OvVg733zjvq1/feBBGRnTn33HPJzMxk3rx5dOnShTPPPJPLL7+czZs3U7VqVcaNG0eLFi1+1TIeNmwY33zzDUuWLOGbb75h6NChDBkyBIAaNWqQnZ3N9OnTGTZsGHXr1mXBggUcfvjhPPbYY5gZL774IldeeSXVq1enS5cuLFmypMQt7ieeeILbb7+dEAJ9+/Zl+PDh5OTkcMEFFzBnzhzMjPPPP58rrriCkSNHMnr0aNLT02nZsiVPPvnkbn2vlMR794aqVT1Zf/utf5w7F0bFhuuHDfOB80qVYJ99YN994aCD4NFYNdl58zyT7rsv1K2b8AFuMw+/alWoU6dsPmcIsGEDrFnjCX3Nmt/eLnh/+fK8+1u2FP5569T5dWIv6thrL7X0RSLTo8dvz51+OlxyibcA+vT57ePnnuvHqlVw6qm/fmz69F0KY9myZcyaNYu0tDTWr1/PzJkzSU9P5/XXX+cvf/kLzz777G+es2jRIt588002bNhAixYtuPjii3+z3nrevHksXLiQhg0b0qVLF9555x2ysrK46KKLmDFjBs2aNWPAgGKndP3Kd999x7XXXsvcuXOpXbs2xxxzDJMnT2bfffdl+fLlLFiwAIC1a9cCcOedd/LVV1+RkZHxy7ndoSTeqZMfRbnkEujc+ddJ/rvv8h6/9lp47TW/nZnpyfzII+GRR/zcjBmeYQ84wLNTCjCDPff0Y7/9Sv/8n3/2lv0PPxR9zJoF33/vvQgFpadDgwb+P1P+5L7PPtC0KTRr5nFVq7bbL1VEktRpp51GWmwCz7p16zjnnHNYvHgxZsa2bdsKfU7fvn3JyMggIyOD+vXrs2LFCho3bvyrazp06PDLubZt27J06VJq1KjB/vvv/8va7AEDBjBmzJgSxTl79mx69Ojxyzj+wIEDmTFjBn/7299YsmQJl112GX379uWYY44BoE2bNgwcOJATTzyRE088sfTfmAKUxHcmK8uPotxzDyxe7Mk996hRI+/xiy6CRYv8dq1asP/+cPzx3sIHeO89z1j77uvZqxyoWjVvVKI4uS3+wpL899/7x2XLYM4c+PFH2LHj189v0CAvqRf82KSJTwQUkVIqruVcrVrxj9etu8st74KqV6/+y+2//e1v9OzZk0mTJrF06VJ6FNZbAGTk+6VPS0tj+/btu3RNWahduzbz58/nlVdeYfTo0Tz99NOMHTuWqVOnMmPGDF544QVuu+02PvnkkyLH/EuifGSNKLVq5UdRJkzwJL9kiR///a/PUgPPYkcd5YPT6enevDzgADjjDDj/fL/m4489M+25Z9xfSqLlb/E3b178tTk5sGIFLF3qx1df5X2cPdu/zfl/F818GkNRSb5x46Sc4iAihVi3bh2NGjUCYPz48WX++Vu0aMGSJUtYunQpTZs25amnnirxczt06MCQIUNYtWoVtWvX5oknnuCyyy5j1apVVKlShVNOOYUWLVpw9tlns2PHDr799lt69uxJ165defLJJ8nOzqZWrVq7HLuSeLwVl+RDgBdeyEvuuR9XrvTH16yBww7z23Xr+lj8oYfC2WdDt26JiT9JpKV5Um7Y0Ec3CsrJ8VGO/Mk99+PMmfD4479uyVeq5J0fTZtCixbQsqUfhxwCjRpp7b5IMvnzn//MOeecw6233krfvn3L/PNXrVqVBx54gN69e1O9enWOOOKIIq+dNm3ar7ron3nmGe6880569uz5y8S2/v37M3/+fM477zx2xP7w3HHHHeTk5HD22Wezbt06QggMGTJktxI4gKVakbSsrKwwZ86cqMNIjE2b4MUX8xL855/DJ5/A8OFw4YV++9hjPbHnP1q1Ul9yAdu2edd8wSS/ZImPdvyUb9PcPfbwZJ4/sbds6R0lWmMv5c1nn33GIYccEnUYkcvOzqZGjRqEELj00ks56KCDuOKKKyKJpbD3xMzmhhB+M7arlngyq1bttzM9Q8hrUlauDEcf7cl81Ki8aeFTp/oM0o8+8pZ+bnJv1qzCTvuuXNlffmF7CoTgnR+ffurHZ5/5x1degfw9d5mZcPDBeUk99+OBB6prXiTVPfTQQzzyyCNs3bqVdu3acdFFF0UdUomoJV5ebN8OX37pCb1XL58JP3o0XHxx3jXVq3srfeJE7zP+4Qcfi69bN7q4k9yaNZ7UcxN77sevv867Jj3dRzryJ/ZDD/WEX07mKko5ppZ48lFLvCJKT/escfDBeecGD/bx84ULPbl/8onfzk3aw4fDP/7hM+a7dPHB5s6dPQNpUBjwwjW535b8Nm70bvjcpP7pp/7tnTQpr6OkalVo2xYOPzzvOOQQJXYRKTtqiVdkc+fCm2/Cu+/CO+/49O8GDXx9lxlMnuzL5Tp29IFi2aktW7zy3fz5/u2dO9frAWVn++NVq/pcxfyJvWVLJXaJjlriyUctcSmZ3CwCPjC8ZImvc89thV9zjXfRV6oEbdp4c7RfPzjuuOhiTnIZGXlTEM4+28/l5Pgqw7lzfc373LleC+j++/3xzMzfttiV2EWkJPRnQpyZr1E/4IC8c7Nnw/vveyt91izPPFu3ehLfscPLLLZv78m9XTvN7ipCWlreSMfAgX5uxw5vsee21ufO9Uq++RN7YS12fYtFJD91p0vJbd/u/cK1avmi7M6d82Z4Va0KHTrA9df7sjcptR078lrsuceHH3pVO/DE3rGjlwjo2tWr+5bDGkCSYFF3p+/OVqTg24NWqVKFzoUUkBg/fjxz5sxhVO5eGClC3ekSH+npnsDBq64sXeo7n8ya5cc77+SVTfvgA7j6aujb15e7tW6tyXI7UamSF55p0QLOOsvP7djhIxpz5/q39O234Y47vIs+d5Sja9e8xN6wYbSvQaS0drYV6c5Mnz6dGjVqFJrEK4KKuWhYyk6jRnDaaV5D/oMPPGkDrF/vTcjrrvNM06SJ15FftSraeFNMpUpeknbAAP8Wz57tu8W9/jrceKMvNBg3ziv1NmrkCw3+8AcYM8ZnzqdYR5sIAHPnzqV79+4cfvjhHHvssXz//fcAjBw5kpYtW9KmTRvOPPNMli5dyujRo7nnnnto27YtM2fOLNHnHzFiBK1bt6Z169b84x//AGDjxo307duXww47jNatW/9SevW666775WuW5p+LRFFLXOLjqKN8Wvby5fDyy1557oUXYORIf3z8eC+T1qePNz3VSi+xGjW8FECsB5Jt23w2/Ntv+/HKK/Dvf/tjder46sGuXf04/HDt6y5FGzrUa0SVpbZtfSVrSYUQuOyyy3juueeoV68eTz31FDfccANjx479zTaetWrVYvDgwaVqvc+dO5dx48bx/vvvE0KgY8eOdO/enSVLltCwYUOmTp0KeL321atXM2nSJBYtWoSZlcnWoWVNSVziq1EjuOACP3bsyKsYN3Wq71py1VVeRq1PHzjxRE/+UiqVK+dttjd0qLe+//tfrxmfm9iff96vzR1Xz03qRx4JNWtGG79Iflu2bGHBggUcffTRAOTk5LDPPvsAZbON59tvv81JJ530yy5pJ598MjNnzqR3795cddVVXHvttfTr149u3bqxfft2MjMzueCCC+jXrx/9+vUrmxdZhpTEJXHyl3x95hmfFPfSS95KHzfO16fnJvF//9ubkPvvH02sKczMS8EeeCCcd56f+/FHn7KQm9jvvNPH1c28dX700X507qyy+xVZaVrM8RJCoFWrVrz77ru/eaywbTzLSvPmzfnwww958cUX+etf/0qvXr248cYb+eCDD5g2bRoTJkxg1KhRvPHGG2X2NcuCxsQlOvvt51Xlnn8eVq+G++7z88uX+8DuAQf4uqwrr/RB4Nza8FJq9evDSSfBiBE+dWHdOpg2zcfVMzPh73+H3//eK9T17g133+274GpMXRItIyODlStX/pLEt23bxsKFC3+1jefw4cNZt24d2dnZ7LHHHmzIXcJRAt26dWPy5Mls2rSJjRs3MmnSJLp168Z3331HtWrVOPvss7nmmmv48MMPyc7OZt26dfTp04d77rmH+fPnx+tl7zK1xCU5ZGbmTa1u1MgXUee20h94wGd1PfSQ7962ebPPlFc1lF1Wvbon7d//HoYN8zmI06f7/0qvveYLC8CT/1FH5bXUY1s6i8RNpUqVmDBhAkOGDGHdunVs376doUOH0rx580K38Tz++OM59dRTee6557jvvvvoVmCb5vHjxzN58uRf7r/33nuce+65dOjQAYALL7yQdu3a8corr3DNNddQqVIlKleuzIMPPsiGDRvo378/mzdvJoTAiBEjEvq9KAmtE5fkt3Gjl4ft2tWXuI0aBbfe6lOyzzrL16drYlyZWrbME3rusWKFnz/kkLyk3qOHqvGWB1GvE5ffKs06cXWnS/KrXt3LveauUT/0UE/o//wndOrkW4jddFPeziOy2xo39oJ8jz3mUxXmz4e77vIRkIcfhhNO8I3yunWDm2/2MgG5JQJEJHGUxCX1dO/uM9tXrPAJcc2awVtv5U2cmzTJx9WlTJj5Uv+rrvIRjjVr4I03vLT+li2exLt08eVs/ft7R8nixVFHLVIxxK073czGAv2AH0MIrYu57gjgXeDMEMKEnX1edadLobZt87VW69ZBvXreLOzRw7vbTznFZ2xJXPz0kyf1117z46uv/Hzz5t6B0revd5xofXpy+uyzzzj44IMxDUklhRACixYtSoru9PFA7+IuMLM0YDjwahzjkIogd2eQmjVhwQLvXl+2DP74R9h7b4hVX5Kyt9decOqpPrqxZImXiR01ylcHjhrlRWnq1YPTT/dNXlaujDpiyS8zM5PVq1eTavOjyqMQAqtXryYzM7PEz4nrxDYzawpMKaolbmZDgW3AEbHr1BKXshOCFx1//HG49FJfsjZ1qq9RP+ssn5qtGe5xlZ3tS9mmTPFvfe5W9Z06eQu9Xz/vqlcjMDrbtm1j2bJlbN68OepQBP+nqnHjxlQusGVhUS3xyJK4mTUCHgd6AmMpJomb2SBgEECTJk0O/zp35yyR0nrwQa/nvn49NGjgM9wvvNAny0lc7djhJT2nTPFj9mw/37ixJ/N+/fz/qqpVo41TJBklYxJ/Brg7hPCemY1HLXFJlM2bff35f/7j2aRxY5+JVamSt97VLEyIH37wt2HKFB9Lz872BP773+eNpe+7b9RRiiSHZEziXwG5fy3rApuAQSGEyQWvzU9JXMrU6tU+kHvEEZ7cO3WCk0+GQYN8LF0SYssWmDHDE/oLL+RNjjvssLxW+hFHQFpatHGKRCXpkniB68ajlrhELXci3Msv+0S5U06BP/3JC4qrdZ4wIcCiRXnj6G+/7XXe69Xz9eknneQFZ1TjXSqShCdxM3sC6IG3slcANwGVAUIIowtcOx4lcUkWixf72PnYsb5k7f33vSqcRGLNGt9e9fnnPamvX++V4vr08U6T445T5Tgp/yJpiceDkrgkzMaNnjnOPNNb4jfc4F3uF1/sW4RJwm3Z4mvSJ02CyZN9uVpGhpeBPekkb6nXrRt1lCJlT2VXRUqrenUYMCCvK/3HH2HkSC/z2qePNwtV6jWhMjK85T1mjC9XmzHD/6f65BPfsr5BA+jZ0zfE+/bbqKMViT+1xEVK47vvfDe10aN9evWf/wzDh0cdVYUXgi9fmzjRW+kLF/r5I47wFvpJJ/mutiKpSt3pImVp61bPFocd5tnh/fc9uV96KbRrF3V0Fd4XX/jbM3Gi758OvgPbSSf5OHr79pqrKKlFSVwknsaN85nsmzb5bPbLL/fZ7VoTFblly3z8fNIk3ycnJweaNMlroXftqrdJkp+SuEi8rV0L48fD/fd7AfGOHeHdd9XkSyKrV/s69IkT4dVXfaJc/freOj/1VN8gT5V4JRkpiYskSk6OZ4k1a7xozI4d8MQTniW0uDlpZGf71qoTJvia9E2bfC36SSf5W9WjR96+OiJRUxIXicobb/hWXg0b+ibcgwZBtWpRRyX5bNrkNX6eecYTena2749+4olw2mleClYJXaKkJWYiUenZE15/3TfYvuIKaNoU7rwTfv456sgkplo171J/4glfSThpEvTuDU8/7R8bNIDzz/da71u3Rh2tSB61xEUS6Z134Lbb4NNPvTJc5cre/a6ZVUlp82YfO58wAZ57zqvF1awJ/ft7C/3oozVCIomh7nSRZLJ2LdSq5VnisMO83/bKK73JJ0lpyxbvUHnmGU/oa9fCnnt6lbhTT4Vjj4XMzKijlPJK3ekiyaRWLf+4YQMcfjjcdZd3sw8ZolJjSSojw7dHHT8eVqzwrvVTTvHCfSee6JPizjrLu+I1UiKJopa4SDJYvNjHyR991JekzZ/v1Ukk6W3b5nMXJ0zwBL56tVfsPeEEOP10H1NXC112l7rTRVLB11/77Kprr/Vk/uST3t2uhJ4Stm2D6dO9y33iRE/oe+yRl9CPPVZj6LJrlMRFUs2WLV5abOVK77e94QZo2zbqqKSEtm2DN9/0Ge65ZQP23NMnxZ1+uibFSeloTFwk1WRk+E4ef/mLT5Fu1863RV26NOrIpAQqV4ZjjoGHH/Yx9Jde8v/FXngBjj/e5zCee66WrcnuUUtcJBWsXQt33w333gtvvw1t2kQdkeyirVt9lvvTT3tN93XroHZtrxR3+ukqLCOFU3e6SHmwfr33yQIMHgz77ONL0/bYI9q4ZJds2QKvveYJPXcd+l57eeGZ00/3OkGq5S6g7nSR8iE3gefkwE8/wbBhcOCB8MADPggrKSUjA/r180UJK1Z4Ij/uOJ/PeMwx/j/aRRfBtGmwfXvU0UoyUhIXSUVpad58e+8938/80kuhZUtQL1XKysz0WeyPPZZX+vXoo+E//4GjjvLS+4MHe1e8ErrkUhIXSWUdO/qapqlToW5daNzYz2dnRxqW7J6qVb2AzOOP++KEZ5/1sfLHHvPEvvfe8Mc/wiuvqAOmotOYuEh5EwJ06+bj5Hfe6evMpVz4+WffbW3CBJ/lvmGDT4rL3W2tVy+oUiXqKCUeNCYuUlHs2OFTnd9/35el/c//aFlaOVG1qr+1//mPd7k/95yPqT/7LPTpA/Xrwznn+HaqW7ZEHa0kglriIuXV2rXeEr/3Xk/sL73kfbJS7hS1Ocvxx3sLXZuzpD4tMROpqJYtgxEjfAvUqlVhwQJo1swLfEu5s3Wrz2afMMHXof/0E9So4Qn91FO9lnu1alFHKaWlJC4iPq25eXPfAvXmm+G887QQuRzLLf06YUJeLffq1X03tlNP9S54/S+XGhI+Jm5mY83sRzNbUMTj/c3sYzP7yMzmmFnXeMUiIjHp6b4ouVkzGDQIWrf2me1SLuWWfh0zBn74wbvczz7bE/vpp/v2qaec4rPe166NOlrZFXFriZvZ74Bs4NEQQutCHq8BbAwhBDNrAzwdQjh4Z59XLXGRMhACPP+875b2+ef+171Xr6ijkgTJyYEZM/K2T/3+e///rlcvnzjXv78vY5PkkfCWeAhhBvBTMY9nh7z/IKoDqdWvL5LKzPwv9ccfe8s8d8LbO+/Apk3RxiZxl5bmJV3vv9+nTLz7LlxxBXz5pReUadjQVymOGAFffRV1tFKcuI6Jm1lTYEphLfHY4ycBdwD1gb4hhHeLuG4QMAigSZMmh3/99ddxiVekQlu/HvbdF2rV8r/eJ5/syV4qjBB83uPEiX58/LGfb9vWfxxOPtkLA+rHIvEimdi2sySe77rfATeGEI7a2edUd7pIHM2cCX/6k//1PuooGDkSDjkk6qgkIv/9r3e3T5oEs2b5uebNvcv95JMhKwsqqdpIQiR1sZdY1/v+ZlY36lhEKrRu3WDuXLjvPq/Dfthh8O23UUclETngALj6ah9lWb7c99nZbz/fFbdjR2jSBC67zCfKqZ57NCJL4mZ2oJl3yphZeyADWB1VPCISk57urfHPP4cHH/QudoDZs72/VSqkhg3h4ovh1Ve9Wtyjj8IRR8DDD/uUir33hvPP92pxmzdHHW3FEc/Z6U8APYC6wArgJqAyQAhhtJldC/wB2Ab8DFwTQnh7Z59X3ekiEfj0Uzj0UOjcGUaNUj12+cXGjb4Ry8SJnsDXrfO150cf7QVm+vTRTPeyoGIvIrLrduyAsWPhuutgzRq45BK45RbffUMkZutW71qfPNnLD+SOxHTo4DXe+/XzSXKaGFd6SuIisvt++gluvNG72Rs1gsWLISMj6qgkCYXg8yOnTPEd1z74wM81apSX0Hv18krAsnNK4iJSdj76CObP9y2zwMfPW7SINiZJaitW+B48U6Z493t2tifwXr28271vX0/wUjglcRGJj6lT/a/wBRfA7bd7LU+RYmzZ4hXjXnjBj9ydctu18x+lfv3g8MO1fC2/pF5iJiIprFs3uPJKGD/eFxHff7/WG0mxMjJ84tvIkbBkCSxc6LvmVq8Ot97qY+iNGsGFF/r4enZ21BEnL7XERaRsfPqpLxp+4w3/C/3qq1FHJClo9eq8bveXX/bZ7hkZXia2d2//0TrkkIo3OU7d6SISfyH4rhppaV7SKyfHFw1rv0vZBdu2wdtve0KfMgW++MLPN2rku7Mdc4wXFqxbAcqEKYmLSOLdey/cc4/PZj/uuKijkRT39dfw2mveyfP6677a0Qzat89L6kceWT4XTGhMXEQSLysLqlXzih9nneVTlEV20X77+Tj500/DypXw/vterqBaNfj7373Lfa+9fKb7vffCZ5+V/yKDaomLSHxt2QLDh8Ntt3m3+sMPe1e7SBlavx6mT89rqed2vTdunNdK79Urdbve1Z0uItFatMg3q775ZujePepopJxbuvTXXe9r1/62671zZ6hSJepIS0ZJXESiF0LetOLrr4caNeCaa1LnL6mkpJwc35Tv1Vf9ePddP1e9uq+Q7NHD/688/HCoXDnqaAunJC4iyWPHDhg4EJ58Elq3hocegk6doo5KKojcrvdXXvFa75995uerV4cuXfKSelZW8vx/qSQuIsnnhRd8M5Xly/3j7bfDnntGHZVUMCtWeAW5t97y5L5woZ+vVs273HOT+hFHRDfzXUlcRJLThg3w17/6hLd587zqm0iEVq7MS+pvveUbuYDXej/yyLyk3rFj4pK6kriIJLcff4T69f323XfDgAHQsGG0MYngVeRmzvRW+ltv+d4/IUBmpo8C5Sb1Tp38XDwoiYtIavjySzj0UB+MHD4cBg3SThiSVNas+XVS/+gjn+aRkeGt8x49YOhQqF277L6mkriIpI4vv/TlaNOm+UyjMWOgZcuooxIp1Nq1Xh42/5j6qlU+pl5WlMRFJLWEAI8+6jukZWbCV18lz1RhkWJs2lS2CRxUdlVEUo0ZnHOOF4l58klP4Dk58MknUUcmUqyyTuDFURIXkeRWr55X5AAYPRratYO//MXLuYpUcEriIpI6Bg701vkdd3j9zNmzo45IJFJK4iKSOmrVgn/9C156yctudeoE//hH1FGJRKZESdzMqptZpdjt5mZ2gpklaYVZESn3eveGBQvg/PO94DWU/z0nRQqRXsLrZgDdzKw28CowGzgDGBivwEREilWzptdcz3XNNT4Z7pZbvLSWSAVQ0u50CyFsAk4GHgghnAa0il9YIiKlEIKv67nrLmjbFmbNijoikYQocRI3syPxlvfU2Lm0nTxhrJn9aGYLinh8oJl9bGafmNksMzus5GGLiORjBg884BtIb9kCXbvCVVd5Yhcpx0qaxIcC1wOTQggLzWx/4M2dPGc80LuYx78CuocQDgX+FxhTwlhERAp31FG+jnzwYHjwQVi2LBzji8gAABTQSURBVOqIROKq1BXbYhPcaoQQ1pfg2qbAlBBC651cVxtYEEJotLPPqYptIlIi333nG6iEAOPHwxlnJLYKh0gZ2q2KbWb2uJntaWbVgQXAp2Z2TRnGdwHwUjFff5CZzTGzOStXrizDLysi5VbuDmjz5vks9jZtfNcKkXKkpN3pLWMt7xPxZNsM+J+yCMDMeuJJ/NqirgkhjAkhZIUQsurVq1cWX1ZEKor27eGNN3ybqe7d4fLLYePGqKMSKRMlTeKVY+vCTwSeDyFsA3Z7UaaZtQEeBvqHEFbv7ucTESlUz57w8cdw6aUwciT06qV15VIulHSd+D+BpcB8YIaZ7QfsdEy8OGbWBJgI/E8I4Yvd+VwiIjtVowbcdx+ceipkZ/uM9pwc+Plnf0wkBe3yVqRmlh5C2F7M408APYC6wArgJqAyQAhhtJk9DJwCfB17yvbCBu0L0sQ2ESkzd9/tS9PGj8/bZEUkCe3uxLaaZjYid3KZmd0NVC/uOSGEASGEfUIIlUMIjUMI/wohjA4hjI49fmEIoXYIoW3s2GkCFxEpUx06+Mfu3eHqq2Hz5mjjESmlko6JjwU2AKfHjvXAuHgFJSKSEN26wfz5cNFF3ipv3x4+/DDqqERKrKRJ/IAQwk0hhCWx42Zg/3gGJiKSEDVqeGGYl1/2Weuq8iYppKRJ/Gcz65p7x8y6AD/HJyQRkQgceywsXuwlW8FnsS8otGq0SNIoaRIfDNxvZkvNbCkwCrgoblGJiEShShX/uG4d3HGHb3M6fLjPYhdJQiVK4iGE+SGEw4A2QJsQQjvg93GNTEQkKjVr+rry44+H667zsfPFi6OOSuQ3StoSByCEsD5fzfQr4xCPiEhyqFcPnnkGHn8cFi2CI4/09eUiSaSkxV4KY2UWhYhIMjKDAQN8CdqcOT4JLgRYuRLq1486OpHStcQLUM1CEakYGjaEE07w2089BQceCP/6l0q3SuSKTeJmtsHM1hdybAAaJihGEZHk0amTT3i78EIfM//++6gjkgqs2CQeQtgjhLBnIcceIYTd6YoXEUlNTZvCtGlw773+sVUrmDgx6qikgtqd7nQRkYqpUiUYMgQ++giaN9cSNImMWtMiIruqRQt45x1IS/P7Dz0EDRrkjZ+LxJla4iIiuyM3ge/Y4ZPd+veH887zgjEicaYkLiJSFipVghkz4IYb4N//hjZt4I03oo5KyjklcRGRslKlCtx6q3exZ2bC0UfDkiVRRyXlmJK4iEhZ69gR5s3zWev7xzZ8XLYs2pikXFISFxGJh2rVfHwcYOZMT+Z//Sts3RptXFKuKImLiMRbmzYwcCDcdpu30rXFqZQRJXERkXirWRPGjYPJk+G777zi2733Rh2VlANK4iIiidK/v7fC+/XzJWkiu0nFXkREEqlePZgwIe/+hAmwahVcdJHvmiZSCmqJi4gkmllewp4wAS6+GPr0geXLo41LUo6SuIhIlB5/HEaNgrfegtat/b62OJUSUhIXEYlSpUpw6aUwfz4ccojPYn/nnaijkhShJC4ikgwOOsjXk0+aBF27+rlvvok2Jkl6cUviZjbWzH40s0IXRJrZwWb2rpltMbOr4xWHiEjKSEuDE0/0259/7ruk/fGPsGFDtHFJ0opnS3w80LuYx38ChgB3xTEGEZHU1LQpXH45jB3rxWLeeivqiCQJxS2JhxBm4Im6qMd/DCHMBrbFKwYRkZSVkQF33uk7o6WlQc+ecNVVmvQmv5ISY+JmNsjM5pjZnJUrV0YdjohI4nTp4pPeLrnEJ8FpLbnkkxJJPIQwJoSQFULIqlevXtThiIgkVvXqvgzt//7P78+YAdddB1u2RBuXRC4lkriIiJDXCp82DYYP9xrsc+dGG5NESklcRCTV3HwzTJ0Ka9b4rmg33aQtTiuouNVON7MngB5AXTNbBtwEVAYIIYw2s72BOcCewA4zGwq0DCGsj1dMIiLlRp8+vpnKkCFwyy3QrBmce27UUUmCWUixmY5ZWVlhzpw5UYchIpI83nwTunf3iW8LF/r68nTtb1WemNncEEJWwfPqThcRSXU9e3oCX7MGfvc7n9G+aFHUUUkCKImLiJQXtWvDAw/Al19Cu3YwYgTk5EQdlcSRkriISHlyxhnepX7MMV4cpkcP2Lgx6qgkTpTERUTKm733hsmT4ZFHoGVLX2cu5ZKSuIhIeWQGf/gD/POffv+zz6BvX/j662jjkjKlJC4iUhF88YVXejv0UHj4YdVgLyeUxEVEKoL+/eGTTyAry7c37dsXvvsu6qhkNymJi4hUFE2bwuuvw8iRMH2612OXlKZqACIiFUmlSnDZZXDccdCokZ/78EOoXx8aN442Nik1tcRFRCqiAw+EqlV9bPzcc6FVK42VpyAlcRGRiswMJk2C9u19rPyYY2Dp0qijkhJSEhcRqegOOMC3N33wQXjvPWjd2ifBSdJTEhcRER8rHzzYq71deql3rwNs2RJtXFIsJXEREcnTpAkMH+5J/YcfvJWuGuxJS0lcRESK1r6912Dv0gU+/TTqaKQAJXERESnc3nvDc8/B44/n7Yx2++2awZ5ElMRFRKRoZjBggI+V51Z9M4s6KolRsRcREdm5Bg3g6adh61a//+mn8MwzcP31UKVKtLFVYGqJi4hIyeUm7EmTYNgwr8U+d26kIVVkSuIiIlJ6N9wAzz8Pq1dDx47eIt+8OeqoKhwlcRER2TXHH+9j5eeeC3feCffeG3VEFY6SuIiI7Lpatbzm+htvwJAhfu7TTyE7O9q4KgglcRER2X09e/qGKtu3+yz21q3hxRejjqrcUxIXEZGyk54O48dDtWrQt68vT1uxIuqoyi0lcRERKVtdusC8eXDLLTBxIhx8MCxaFHVU5VLckriZjTWzH81sQRGPm5mNNLMvzexjM2sfr1hERCTBMjLgb3+Djz/2iW/Nm/v5n3+ONKzyJp4t8fFA72IePw44KHYMAh6MYywiIhKFFi3gnnvyNlTZf3/43//NKxojuyVuSTyEMAP4qZhL+gOPBvceUMvM9olXPCIiErG0NOjeHW680euwz5oVdUQpL8ox8UbAt/nuL4ud+w0zG2Rmc8xszsqVKxMSnIiIlLF69eDJJ2HKFF+C1qULXHKJtjndDSkxsS2EMCaEkBVCyKpXr17U4YiIyO7o29eLxFxxBWzY4C102SVRJvHlwL757jeOnRMRkfKuRg0YMQIefdTvL1gAp5wC335b/PPkV6JM4s8Df4jNUu8ErAshfB9hPCIikmi525ouXAgvvQQtW8J996mLvYTiucTsCeBdoIWZLTOzC8xssJkNjl3yIrAE+BJ4CLgkXrGIiEiSO+MMT+SdO3v51s6dfXmaFCtu+4mHEAbs5PEAXBqvry8iIimmWTN4+WV4/HEYOtQnwbVpE3VUSS1uSVxERKTUzGDgQOjd20u3AsyYAdu2Qa9e0caWhFJidrqIiFQwder4hioAt98ORx0FZ5/tBWPkF0riIiKS3CZNgr/+FZ55xivA3Xef75YmSuIiIpLkqlb1Uq0LFkCnTj7x7dlno44qKSiJi4hIajjoIJ/49vLLcNppfu7NN2HVqmjjipCSuIiIpA4zOPZY31Bl82Zfmta8Ofzzn7BjR9TRJZySuIiIpKbMTG+Jt2kDgwd7V/vcuVFHlVBK4iIikrpatfJE/thj8M030KEDfP551FEljJK4iIiktty15YsWwcMP+wx28FZ5Oe9iVxIXEZHyoVYtOO88v714sXev/+535bp8q5K4iIiUPwccAKNHe+u8fXvf9nT9+qijKnNK4iIiUv5UqgQXXODj4xdcAPfe6xPgNm+OOrIypSQuIiLlV506vvzsvffguut8RjvA8uXRxlVGlMRFRKT869DBl6EBvPqq75h2/fWwcWO0ce0mJXEREalY2rb12ex33gkHH+xbn4YQdVS7RElcREQqlvr1Ydw4ePttvz1wIJx4YtRR7RLtJy4iIhVTly4wezaMHw9Vqvi5nBxYuRL23jvS0EpKLXEREam4KlWC88/3vcoB/vUv32jljjtSYia7kriIiEiuXr3gqKPgL3+Bli1h4sSkHi9XEhcREcl1wAEwaRK8/jpUrw6nnAIXXRR1VEXSmLiIiEhBvXrBvHnw0EO+1SlAdjb8/DPUqxdtbPmoJS4iIlKY9HS4+GJP6AC33ebj5SNGwNat0cYWoyQuIiJSEn/4Axx5JFx1FRx6KEydGvl4uZK4iIhISRxyCLz0kidvgH794KabIg1JSVxERKQ0+vSBTz7xbvUzzvBzK1bATz8lPJS4JnEz621mn5vZl2Z2XSGP72dm08zsYzObbmaN4xmPiIhImahSxbc3bdXK7w8d6uPl998P27cnLIy4JXEzSwPuB44DWgIDzKxlgcvuAh4NIbQBbgHuiFc8IiIicXP99XDYYfCnP8FddyXsy8azJd4B+DKEsCSEsBV4Euhf4JqWwBux228W8riIiEjya9MGpk3zNeYXX5ywLxvPJN4I+Dbf/WWxc/nNB06O3T4J2MPM6hT8RGY2yMzmmNmclStXxiVYERGR3WLmG6nUrJmwLxn1xLarge5mNg/oDiwHcgpeFEIYE0LICiFk1UuiRfYiIiJRimfFtuXAvvnuN46d+0UI4TtiLXEzqwGcEkJYG8eYREREyo14tsRnAweZWTMzqwKcCTyf/wIzq2tmuTFcD4yNYzwiIiLlStySeAhhO/An4BXgM+DpEMJCM7vFzE6IXdYD+NzMvgAaALfFKx4REZHyxkISb7FWmKysrDBnzpyowxAREUkYM5sbQsgqeD7qiW0iIiKyi5TERUREUpSSuIiISIpKuTFxM1sJfB11HGWsLrAq6iDKmF5T6iiPr6s8viYon69Lr6lk9gsh/KZQSsol8fLIzOYUNmEhlek1pY7y+LrK42uC8vm69Jp2j7rTRUREUpSSuIiISIpSEk8OY6IOIA70mlJHeXxd5fE1Qfl8XXpNu0Fj4iIiIilKLXEREZEUpSQuIiKSopTEE8DM9jWzN83sUzNbaGaXF3JNDzNbZ2YfxY4bo4i1tMxsqZl9Eov5N0XtzY00sy/N7GMzax9FnCVlZi3yvQcfmdl6Mxta4JqUeK/MbKyZ/WhmC/Kd28vMXjOzxbGPtYt47jmxaxab2TmJi7p4Rbymv5vZotjP1yQzq1XEc4v9WY1SEa9rmJktz/dz1qeI5/Y2s89jv2PXJS7q4hXxmp7K93qWmtlHRTw3Kd+rov6WR/p7FULQEecD2AdoH7u9B/AF0LLANT2AKVHHuguvbSlQt5jH+wAvAQZ0At6POuZSvLY04Ae8yELKvVfA74D2wIJ85/4PuC52+zpgeCHP2wtYEvtYO3a7dtSvp5jXdAyQHrs9vLDXFHus2J/VJHxdw4Crd/K8NOC/wP5AFWB+wb8tyfSaCjx+N3BjKr1XRf0tj/L3Si3xBAghfB9C+DB2ewO+NWujaKNKmP7Ao8G9B9Qys32iDqqEegH/DSGkZIXAEMIM4KcCp/sDj8RuPwKcWMhTjwVeCyH8FEJYA7wG9I5boKVQ2GsKIbwafOtjgPeAxgkPbDcV8V6VRAfgyxDCkhDCVuBJ/D2OXHGvycwMOB14IqFB7aZi/pZH9nulJJ5gZtYUaAe8X8jDR5rZfDN7ycxaJTSwXReAV81srpkNKuTxRsC3+e4vI3X+gTmTov/IpOJ7BdAghPB97PYPQINCrknl9+x8vOenMDv7WU1Gf4oNE4wtoos2Vd+rbsCKEMLiIh5P+veqwN/yyH6vlMQTyMxqAM8CQ0MI6ws8/CHebXsYcB8wOdHx7aKuIYT2wHHApWb2u6gDKgtmVgU4AXimkIdT9b36leB9fOVmjamZ3QBsB/5TxCWp9rP6IHAA0Bb4Hu9+Li8GUHwrPKnfq+L+lif690pJPEHMrDL+pv8nhDCx4OMhhPUhhOzY7ReBymZWN8FhlloIYXns44/AJLx7L7/lwL757jeOnUt2xwEfhhBWFHwgVd+rmBW5wxmxjz8Wck3KvWdmdi7QDxgY+yP6GyX4WU0qIYQVIYScEMIO4CEKjzcV36t04GTgqaKuSeb3qoi/5ZH9XimJJ0Bs/OdfwGchhBFFXLN37DrMrAP+3qxOXJSlZ2bVzWyP3Nv4BKMFBS57HvhDbJZ6J2Bdvm6nZFZkSyEV36t8ngdyZ8WeAzxXyDWvAMeYWe1YF+4xsXNJycx6A38GTgghbCrimpL8rCaVAnNHTqLweGcDB5lZs1jv0Zn4e5zMjgIWhRCWFfZgMr9Xxfwtj+73KurZfhXhALri3SsfAx/Fjj7AYGBw7Jo/AQvx2aXvAZ2jjrsEr2v/WLzzY7HfEDuf/3UZcD8+g/YTICvquEvwuqrjSblmvnMp917h/4R8D2zDx98uAOoA04DFwOvAXrFrs4CH8z33fODL2HFe1K9lJ6/pS3ysMfd3a3Ts2obAi8X9rCbLUcTr+nfsd+ZjPEnsU/B1xe73wWdJ/zeZXldhryl2fnzu71K+a1PivSrmb3lkv1cquyoiIpKi1J0uIiKSopTERUREUpSSuIiISIpSEhcREUlRSuIiIiIpSklcpIIxsxz79U5tZbbzlZk1zb9rlYjEV3rUAYhIwv0cQmgbdRAisvvUEhcR4Jc9nP8vto/zB2Z2YOx8UzN7I7YRxzQzaxI738B8/+75saNz7FOlmdlDsf2WXzWzqpG9KJFyTklcpOKpWqA7/Yx8j60LIRwKjAL+ETt3H/BICKENvrnIyNj5kcBbwTeCaY9X1wI4CLg/hNAKWAucEufXI1JhqWKbSAVjZtkhhBqFnF8K/D6EsCS2ycMPIYQ6ZrYKL/m5LXb++xBCXTNbCTQOIWzJ9zma4nsmHxS7fy1QOYRwa/xfmUjFo5a4iOQXirhdGlvy3c5Bc29E4kZJXETyOyPfx3djt2fhu2MBDARmxm5PAy4GMLM0M6uZqCBFxOk/ZJGKp6qZfZTv/sshhNxlZrXN7GO8NT0gdu4yYJyZXQOsBM6Lnb8cGGNmF+At7ovxXatEJEE0Ji4iwC9j4lkhhFVRxyIiJaPudBERkRSllriIiEiKUktcREQkRSmJi4iIpCglcRERkRSlJC4iIpKilMRFRERS1P8D08uSsq7VO1EAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MRaPd129QrpR"
      },
      "source": [
        "loss, accuracy= model.evaluate(X_test_array, y_test,verbose=0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "uUacIlKLRk2e",
        "outputId": "5ab57ee9-ac05-4d23-f0e9-caaa07fc0297"
      },
      "source": [
        "model.evaluate(X_test_array, y_test,verbose=0)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[1.2341417074203491, 0.006071350537240505, 0.199106827378273]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 282
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qYmbP0NDRp9w"
      },
      "source": [
        "y_pred = model.predict(X_test_array)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Uq74oCxUnRSQ"
      },
      "source": [
        "pre_k = []\n",
        "for k in range(2, 31):\n",
        "    pre = precisionatk(y_test, y_pred, k)\n",
        "    pre_k.append(pre)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "FnJpauIhnRAm",
        "outputId": "48f2e6e6-0abd-4395-94e4-b95edde586bb"
      },
      "source": [
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(range(2, 31), pre_k)\n",
        "plt.xlabel('k')\n",
        "plt.ylabel('precision@K')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f4mf24Hhe5rc"
      },
      "source": [
        "SVG(model_to_dot( model,  show_shapes=True, show_layer_names=True, dpi=65).create(prog='dot', format='svg'))"
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}