Copy_of_NN_subtask_A.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of NN_subtask_A.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/viscory/semeval3108/blob/master/Copy_of_NN_subtask_A.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "n5dbbADb-wvZ",
        "colab_type": "code",
        "outputId": "fbfdc00a-feff-41ff-b0b1-489be2e7626e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 80
        }
      },
      "source": [
        "import numpy as np\n",
        "import csv\n",
        "from tqdm import tqdm\n",
        "from sklearn.utils import shuffle\n",
        "import nltk\n",
        "import re\n",
        "import keras\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Input, Dense, Embedding, Convolution1D, MaxPooling1D, MaxPooling2D, Convolution2D, LSTM, Reshape, Dropout, Flatten\n",
        "from keras import regularizers \n",
        "from sklearn.feature_extraction.text import TfidfVectorizer,CountVectorizer\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.preprocessing import OneHotEncoder\n",
        "from keras.optimizers import Adam\n",
        "from tensorflow.python.keras.callbacks import EarlyStopping\n",
        "import seaborn as sn\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.metrics import classification_report"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "display_data",
          "data": {
            "text/html": [
              "<p style=\"color: red;\">\n",
              "The default version of TensorFlow in Colab will soon switch to TensorFlow 2.x.<br>\n",
              "We recommend you <a href=\"https://www.tensorflow.org/guide/migrate\" target=\"_blank\">upgrade</a> now \n",
              "or ensure your notebook will continue to use TensorFlow 1.x via the <code>%tensorflow_version 1.x</code> magic:\n",
              "<a href=\"https://colab.research.google.com/notebooks/tensorflow_version.ipynb\" target=\"_blank\">more info</a>.</p>\n"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bMUkEtUY_BMz",
        "colab_type": "code",
        "outputId": "dd8051a2-57c1-4506-9c19-0b168f051070",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 122
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/gdrive')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3aietf%3awg%3aoauth%3a2.0%3aoob&response_type=code&scope=email%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdocs.test%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive%20https%3a%2f%2fwww.googleapis.com%2fauth%2fdrive.photos.readonly%20https%3a%2f%2fwww.googleapis.com%2fauth%2fpeopleapi.readonly\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U0wDqIgg_K_w",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#declare path\n",
        "path_to_project = '/content/gdrive/My Drive/ESTR project AI/OLIDv1.0/'\n",
        "path_to_data = path_to_project + 'olid-training-v1.0.tsv'"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9Pery0ndB34Y",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#0 - Not offensive\n",
        "#1 - Offensive untargeted\n",
        "#2 - Offensive targeted indiviualds\n",
        "#3 - Offensive targeted groups \n",
        "#4 - Offensive targeted others\n",
        "def get_label(labels):\n",
        "  answer = 0\n",
        "  if labels[1] == 'UNT':\n",
        "    answer = 1\n",
        "  if labels[2] == 'IND':\n",
        "    answer = 2\n",
        "  if labels[2] == 'GRP':\n",
        "    answer = 3\n",
        "  if labels[2] == 'OTH':\n",
        "    answer = 4\n",
        "  return answer\n",
        "\n",
        "def convert(label, sub_task):\n",
        "  if sub_task == 'a':\n",
        "    return int(label > 0)\n",
        "  if sub_task == 'b':\n",
        "    return int(label > 1)\n",
        "  return label - 2\n",
        "\n",
        "def prepare_data(path,sub_task):\n",
        "  data=[]\n",
        "  labels=[]\n",
        "  with open(path,encoding='utf8') as file:\n",
        "    for i,line in enumerate(csv.reader(file,delimiter='\\t')):\n",
        "      if i == 0:\n",
        "        continue\n",
        "      data.append(line[1])\n",
        "      labels.append(convert(get_label(line[-3:]),sub_task))\n",
        "  return data,labels  \n",
        "\n",
        "def get_test_label(label,sub_task):\n",
        "  answer = 0\n",
        "  if sub_task == 'a':\n",
        "    if label == 'OFF':\n",
        "      answer = 1\n",
        "  if sub_task == 'b':\n",
        "    if label == 'TIN':\n",
        "      answer = 1\n",
        "  if sub_task == 'c':\n",
        "    if label == 'GRP':\n",
        "      answer = 1\n",
        "    if label == 'OTH':\n",
        "      answer = 2\n",
        "  return answer\n",
        "def get_test_data(sub_task):\n",
        "  data=[]\n",
        "  labels=[]\n",
        "  path_labels = path_to_project + 'labels-level' + sub_task + '.csv'\n",
        "  path_data = path_to_project + 'testset-level' + sub_task + '.tsv'\n",
        "  with open(path_data,encoding='utf8') as file:\n",
        "    for i,line in enumerate(csv.reader(file,delimiter='\\t')):\n",
        "      if i == 0:\n",
        "        continue\n",
        "      data.append(line[1])\n",
        "  with open(path_labels,encoding='utf8') as file:\n",
        "    for line in csv.reader(file):\n",
        "      labels.append(get_test_label(line[1],sub_task))\n",
        "      #labels.append(line[1])\n",
        "  return data,labels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1e-u1aaLYc27",
        "colab_type": "code",
        "outputId": "c6f0b17a-6cd1-4e14-889d-b281cf03dc22",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 170
        }
      },
      "source": [
        "#preprocessing dependencies\n",
        "nltk.download('punkt')\n",
        "nltk.download('stopwords')\n",
        "nltk.download('averaged_perceptron_tagger')\n",
        "nltk.download('wordnet')\n",
        "#preprocessing\n",
        "def tokenize(data):\n",
        "  answer = []\n",
        "  for text in data:\n",
        "    answer.append(nltk.word_tokenize(text.lower()))\n",
        "  return answer\n",
        "\n",
        "def stop_words(data):\n",
        "  answer = []\n",
        "  bad_words = set(nltk.corpus.stopwords.words('english'))\n",
        "  for text in data:\n",
        "    answer.append([x for x in text if not x in bad_words and not re.match(r\"[^a-zA-Z\\d\\s]+\", x) and x != 'user'])\n",
        "  return answer\n",
        "\n",
        "def get_tag(word):\n",
        "  tag = nltk.pos_tag([word])[0][1][0].upper()\n",
        "  tag_dict = {\"J\": nltk.corpus.wordnet.ADJ,\n",
        "                \"N\": nltk.corpus.wordnet.NOUN,\n",
        "                \"V\": nltk.corpus.wordnet.VERB,\n",
        "                \"R\": nltk.corpus.wordnet.ADV}\n",
        "  return tag_dict.get(tag, nltk.corpus.wordnet.NOUN)\n",
        "  \n",
        "def lemmatize(data):\n",
        "  answer = []      \n",
        "  lemmatizer = nltk.stem.WordNetLemmatizer()\n",
        "  for text in data:\n",
        "    answer.append([lemmatizer.lemmatize(x, pos=get_tag(x)) for x in text])\n",
        "  return answer\n",
        "  \n",
        "#Vectorization\n",
        "def count_vectorizer_len(data):\n",
        "  vectorizer = CountVectorizer()\n",
        "  converted_data = [' '.join(x) for x in data]\n",
        "  vectorizer.fit(converted_data)\n",
        "  return len(vectorizer.vocabulary_.keys())"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[nltk_data] Downloading package punkt to /root/nltk_data...\n",
            "[nltk_data]   Unzipping tokenizers/punkt.zip.\n",
            "[nltk_data] Downloading package stopwords to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/stopwords.zip.\n",
            "[nltk_data] Downloading package averaged_perceptron_tagger to\n",
            "[nltk_data]     /root/nltk_data...\n",
            "[nltk_data]   Unzipping taggers/averaged_perceptron_tagger.zip.\n",
            "[nltk_data] Downloading package wordnet to /root/nltk_data...\n",
            "[nltk_data]   Unzipping corpora/wordnet.zip.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "0cgvEFhxxw9w",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#preparing data to be compatibel with Neural networks\n",
        "def do_one_hot(data):\n",
        "  encoder = OneHotEncoder();\n",
        "  return encoder.fit_transform(np.array(data).reshape(-1,1)).toarray()\n",
        "def do_input_proc(data, len, longest_seq):\n",
        "  converted_data = [keras.preprocessing.text.one_hot(' '.join(x), len) for x in data]\n",
        "  return keras.preprocessing.sequence.pad_sequences(converted_data, maxlen=longest_seq, padding='post')\n",
        "def get_max_len(data):\n",
        "  answer = 0\n",
        "  for x in data:\n",
        "    answer = max(answer, len(x))\n",
        "  return answer"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "App7TmF4MxxX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def plot_confusion(data, labels, model, classes=2):\n",
        "  predicted_labels = model.predict(data, batch_size=64)\n",
        "  mtx = np.zeros(shape=(classes,classes))\n",
        "  for i in range(0,len(predicted_labels)):\n",
        "    actual_label = np.argmax(labels[i])\n",
        "    predicted_label = np.argmax(predicted_labels[i])\n",
        "    mtx[actual_label,predicted_label] += 1\n",
        "  norm = np.zeros(shape=(classes,classes))\n",
        "  for i in range(classes):\n",
        "    for j in range(classes):\n",
        "      norm[i][j] = mtx[i][j] / np.sum(mtx[i])\n",
        "  print(norm)\n",
        "  plt.imshow(mtx,cmap=plt.cm.ocean_r)\n",
        "  plt.title('Confusion matrix')\n",
        "  plt.colorbar()\n",
        "  marks = [x for x in ['NOT', 'OFF']]\n",
        "  plt.xlabel('Predicted Label')\n",
        "  plt.xticks(np.arange(classes), marks)\n",
        "  plt.ylabel('True Label')\n",
        "  plt.yticks(np.arange(classes), marks)\n",
        "  plt.tight_layout()\n",
        "  for (i, j), z in np.ndenumerate(mtx):\n",
        "    plt.text(j, i, int(z), ha='center', va='center')\n",
        "  plt.show()\n",
        "def do_analysis(test_data, test_labels, model):\n",
        "  predicted_labels = model.predict(test_data, batch_size=64)\n",
        "  y_true = np.argmax(test_labels, axis = 1)\n",
        "  y_pred = np.argmax(predicted_labels, axis = 1)\n",
        "  print(classification_report(y_true, y_pred))\n",
        "  plot_confusion(test_data, test_labels, model, 2)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bWzRzfGyBXKp",
        "colab_type": "code",
        "outputId": "ed7aa826-31f4-4b53-d0fe-b2bedbc9f9fc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 241
        }
      },
      "source": [
        "data,labels = prepare_data(path_to_data,'a')\n",
        "test_data,test_labels = get_test_data('a')\n",
        "data = lemmatize(stop_words(tokenize(data)))\n",
        "test_data = lemmatize(stop_words(tokenize(test_data)))\n",
        "vocab_length = count_vectorizer_len(data)\n",
        "longest_seq = max(get_max_len(data),get_max_len(test_data))\n",
        "#convert data to work with NN\n",
        "train_data, val_data, train_labels, val_labels = train_test_split(np.array(data),labels, test_size=0.35,stratify=labels)\n",
        "train_labels = do_one_hot(train_labels)\n",
        "val_labels = do_one_hot(val_labels)\n",
        "test_labels = do_one_hot(test_labels)\n",
        "train_data = do_input_proc(train_data, vocab_length, longest_seq)\n",
        "val_data = do_input_proc(val_data, vocab_length, longest_seq)\n",
        "test_data = do_input_proc(test_data, vocab_length, longest_seq)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/usr/local/lib/python3.6/dist-packages/sklearn/preprocessing/_encoders.py:415: FutureWarning: The handling of integer data will change in version 0.22. Currently, the categories are determined based on the range [0, max(values)], while in the future they will be determined based on the unique values.\n",
            "If you want the future behaviour and silence this warning, you can specify \"categories='auto'\".\n",
            "In case you used a LabelEncoder before this OneHotEncoder to convert the categories to integers, then you can now use the OneHotEncoder directly.\n",
            "  warnings.warn(msg, FutureWarning)\n",
            "/usr/local/lib/python3.6/dist-packages/sklearn/preprocessing/_encoders.py:415: FutureWarning: The handling of integer data will change in version 0.22. Currently, the categories are determined based on the range [0, max(values)], while in the future they will be determined based on the unique values.\n",
            "If you want the future behaviour and silence this warning, you can specify \"categories='auto'\".\n",
            "In case you used a LabelEncoder before this OneHotEncoder to convert the categories to integers, then you can now use the OneHotEncoder directly.\n",
            "  warnings.warn(msg, FutureWarning)\n",
            "/usr/local/lib/python3.6/dist-packages/sklearn/preprocessing/_encoders.py:415: FutureWarning: The handling of integer data will change in version 0.22. Currently, the categories are determined based on the range [0, max(values)], while in the future they will be determined based on the unique values.\n",
            "If you want the future behaviour and silence this warning, you can specify \"categories='auto'\".\n",
            "In case you used a LabelEncoder before this OneHotEncoder to convert the categories to integers, then you can now use the OneHotEncoder directly.\n",
            "  warnings.warn(msg, FutureWarning)\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "W9r-p3QZZiK5",
        "colab_type": "code",
        "outputId": "06327ca2-c462-41dd-8600-9b450a089d24",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 289
        }
      },
      "source": [
        "# lstm model\n",
        "lstm = Sequential()\n",
        "lstm.add(Embedding(vocab_length, 30, input_length=longest_seq))\n",
        "lstm.add(Dropout(0.5))\n",
        "lstm.add(LSTM(100))\n",
        "#lstm.add(Dense(longest_seq, activation='relu', W_regularizer=regularizers.l2(0.90))) , W_regularizer=regularizers.l2(0.1))\n",
        "lstm.add(Dense(train_labels.shape[1], activation='softmax'))\n",
        "madam = Adam(lr=0.008)\n",
        "lstm.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
        "lstm.summary()\n",
        "early_stopping = EarlyStopping(monitor='val_acc', patience=5)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_5\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding_5 (Embedding)      (None, 44, 30)            472380    \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 44, 30)            0         \n",
            "_________________________________________________________________\n",
            "lstm_4 (LSTM)                (None, 100)               52400     \n",
            "_________________________________________________________________\n",
            "dense_7 (Dense)              (None, 2)                 202       \n",
            "=================================================================\n",
            "Total params: 524,982\n",
            "Trainable params: 524,982\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "foB-iGHaIN7w",
        "colab_type": "code",
        "outputId": "c523142e-4bfd-4fa3-ca48-3aa6435a10da",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "lstm.fit(train_data, train_labels, batch_size=64,epochs=5,verbose=1,validation_data=(val_data,val_labels),callbacks=[early_stopping])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 8606 samples, validate on 4634 samples\n",
            "Epoch 1/5\n",
            "8606/8606 [==============================] - 18s 2ms/step - loss: 0.6394 - acc: 0.6678 - val_loss: 0.6357 - val_acc: 0.6677\n",
            "Epoch 2/5\n",
            "8606/8606 [==============================] - 17s 2ms/step - loss: 0.6372 - acc: 0.6677 - val_loss: 0.6370 - val_acc: 0.6677\n",
            "Epoch 3/5\n",
            "8606/8606 [==============================] - 17s 2ms/step - loss: 0.6367 - acc: 0.6677 - val_loss: 0.6353 - val_acc: 0.6677\n",
            "Epoch 4/5\n",
            "8606/8606 [==============================] - 17s 2ms/step - loss: 0.5573 - acc: 0.7264 - val_loss: 0.6228 - val_acc: 0.7268\n",
            "Epoch 5/5\n",
            "8606/8606 [==============================] - 17s 2ms/step - loss: 0.4347 - acc: 0.8136 - val_loss: 0.5415 - val_acc: 0.7393\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f93ed5426d8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 24
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WLxnyIaTJVCs",
        "colab_type": "code",
        "outputId": "61933809-11af-4130-ee59-6903f356db6e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "print(lstm.metrics_names, lstm.evaluate(test_data, test_labels, batch_size=64, verbose=1))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "860/860 [==============================] - 1s 631us/step\n",
            "['loss', 'acc'] [0.4850301146507263, 0.7709302320036777]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vDYlvIJ8Orwh",
        "colab_type": "code",
        "outputId": "2ee86757-a09a-4914-a1db-1ce76f9afb59",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 484
        }
      },
      "source": [
        "do_analysis(test_data, test_labels, lstm)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.84      0.84      0.84       620\n",
            "           1       0.59      0.60      0.59       240\n",
            "\n",
            "    accuracy                           0.77       860\n",
            "   macro avg       0.72      0.72      0.72       860\n",
            "weighted avg       0.77      0.77      0.77       860\n",
            "\n",
            "[[0.83709677 0.16290323]\n",
            " [0.4        0.6       ]]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAU0AAAEYCAYAAADYs6SAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nSUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gKY7TG9sH7HZ",
        "colab_type": "code",
        "outputId": "997162ae-1a8d-43f5-f44d-2b0979c90dcf",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 323
        }
      },
      "source": [
        "# lstm model\n",
        "lstm_complex = Sequential()\n",
        "lstm_complex.add(Embedding(vocab_length, 30, input_length=longest_seq))\n",
        "#lstm_complex.add(Dropout(0.4))\n",
        "lstm_complex.add(LSTM(50))\n",
        "lstm_complex.add(Dense(longest_seq, activation='relu'))\n",
        "lstm_complex.add(Dropout(0.75))\n",
        "lstm_complex.add(Dense(train_labels.shape[1], activation='sigmoid'))\n",
        "#madam = Adam(lr=0.008)\n",
        "lstm_complex.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
        "lstm_complex.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_12\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding_12 (Embedding)     (None, 44, 30)            472380    \n",
            "_________________________________________________________________\n",
            "lstm_11 (LSTM)               (None, 50)                16200     \n",
            "_________________________________________________________________\n",
            "dense_20 (Dense)             (None, 44)                2244      \n",
            "_________________________________________________________________\n",
            "dropout_18 (Dropout)         (None, 44)                0         \n",
            "_________________________________________________________________\n",
            "dense_21 (Dense)             (None, 2)                 90        \n",
            "=================================================================\n",
            "Total params: 490,914\n",
            "Trainable params: 490,914\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bIupfnzheuKB",
        "colab_type": "code",
        "outputId": "38fc0c0b-0937-466e-9c61-af09fd2362f6",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 221
        }
      },
      "source": [
        "early_stopping = EarlyStopping(monitor='val_loss', patience=2)\n",
        "lstm_complex.fit(train_data, train_labels, batch_size=64,epochs=8,verbose=1,validation_data=(val_data,val_labels),callbacks=[early_stopping])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 8606 samples, validate on 4634 samples\n",
            "Epoch 1/8\n",
            "8606/8606 [==============================] - 20s 2ms/step - loss: 0.6506 - acc: 0.6666 - val_loss: 0.6375 - val_acc: 0.6677\n",
            "Epoch 2/8\n",
            "8606/8606 [==============================] - 16s 2ms/step - loss: 0.6018 - acc: 0.7011 - val_loss: 0.5530 - val_acc: 0.7128\n",
            "Epoch 3/8\n",
            "8606/8606 [==============================] - 16s 2ms/step - loss: 0.4582 - acc: 0.8144 - val_loss: 0.5432 - val_acc: 0.7393\n",
            "Epoch 4/8\n",
            "8606/8606 [==============================] - 16s 2ms/step - loss: 0.3657 - acc: 0.8695 - val_loss: 0.5896 - val_acc: 0.7391\n",
            "Epoch 5/8\n",
            "8606/8606 [==============================] - 16s 2ms/step - loss: 0.3093 - acc: 0.8960 - val_loss: 0.5961 - val_acc: 0.7352\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f93eb405f98>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "7SCbTzzcgDwt",
        "colab_type": "code",
        "outputId": "c05c3b74-796f-4c8e-9477-af73670f6139",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "print(lstm_complex.metrics_names, lstm_complex.evaluate(test_data, test_labels, batch_size=64, verbose=1))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "860/860 [==============================] - 1s 626us/step\n",
            "['loss', 'acc'] [0.5510285197302353, 0.7697674426921578]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xf7tFaXXg2h-",
        "colab_type": "code",
        "outputId": "8341f16f-6804-407e-f537-b50f2e90a309",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 484
        }
      },
      "source": [
        "do_analysis(test_data, test_labels, lstm_complex)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.82      0.87      0.85       620\n",
            "           1       0.60      0.51      0.55       240\n",
            "\n",
            "    accuracy                           0.77       860\n",
            "   macro avg       0.71      0.69      0.70       860\n",
            "weighted avg       0.76      0.77      0.76       860\n",
            "\n",
            "[[0.87096774 0.12903226]\n",
            " [0.49166667 0.50833333]]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAU0AAAEYCAYAAADYs6SAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nmmZmBfx/tqiDAwvfPKAAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KqPKueZxIS0b",
        "colab_type": "code",
        "outputId": "9558bda2-63a3-4555-c038-7b48862001bb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 649
        }
      },
      "source": [
        "cnn = Sequential()\n",
        "cnn.add(Embedding(vocab_length, 30, input_length=longest_seq))\n",
        "cnn.add(Reshape((30,longest_seq,1)))\n",
        "cnn.add(Convolution2D(32,(1,5),activation=\"relu\"))\n",
        "cnn.add(Dropout(0.9))\n",
        "cnn.add(Convolution2D(16,(2,3),activation=\"relu\"))\n",
        "cnn.add(Dropout(0.8))\n",
        "cnn.add(Convolution2D(16,(2,2),activation=\"relu\"))\n",
        "cnn.add(Dropout(0.7))\n",
        "cnn.add(Flatten())\n",
        "cnn.add(Dense(train_labels.shape[1],activation='softmax'))\n",
        "cnn.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
        "cnn.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:148: The name tf.placeholder_with_default is deprecated. Please use tf.compat.v1.placeholder_with_default instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3733: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "WARNING:tensorflow:Large dropout rate: 0.9 (>0.5). In TensorFlow 2.x, dropout() uses dropout rate instead of keep_prob. Please ensure that this is intended.\n",
            "WARNING:tensorflow:Large dropout rate: 0.8 (>0.5). In TensorFlow 2.x, dropout() uses dropout rate instead of keep_prob. Please ensure that this is intended.\n",
            "WARNING:tensorflow:Large dropout rate: 0.7 (>0.5). In TensorFlow 2.x, dropout() uses dropout rate instead of keep_prob. Please ensure that this is intended.\n",
            "Model: \"sequential_3\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding_3 (Embedding)      (None, 44, 30)            472380    \n",
            "_________________________________________________________________\n",
            "reshape_1 (Reshape)          (None, 30, 44, 1)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 30, 40, 32)        192       \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 30, 40, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 29, 38, 16)        3088      \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 29, 38, 16)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 28, 37, 16)        1040      \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 28, 37, 16)        0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 16576)             0         \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 2)                 33154     \n",
            "=================================================================\n",
            "Total params: 509,854\n",
            "Trainable params: 509,854\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QYygOoUcIYY7",
        "colab_type": "code",
        "outputId": "888b71b8-f6a6-462a-8dc0-39f3a330fc5e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "source": [
        "early_stopping = EarlyStopping(monitor='val_acc', patience=5)\n",
        "cnn.fit(train_data, train_labels, batch_size=64,epochs=60,verbose=1,validation_data=(val_data,val_labels),callbacks=[early_stopping])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 8606 samples, validate on 4634 samples\n",
            "Epoch 1/60\n",
            "8606/8606 [==============================] - 6s 672us/step - loss: 0.6398 - acc: 0.6657 - val_loss: 0.6783 - val_acc: 0.6677\n",
            "Epoch 2/60\n",
            "8606/8606 [==============================] - 2s 277us/step - loss: 0.6321 - acc: 0.6677 - val_loss: 0.6856 - val_acc: 0.6677\n",
            "Epoch 3/60\n",
            "8606/8606 [==============================] - 2s 279us/step - loss: 0.6029 - acc: 0.6770 - val_loss: 0.6710 - val_acc: 0.6474\n",
            "Epoch 4/60\n",
            "8606/8606 [==============================] - 2s 278us/step - loss: 0.5274 - acc: 0.7432 - val_loss: 0.6434 - val_acc: 0.6675\n",
            "Epoch 5/60\n",
            "8606/8606 [==============================] - 2s 277us/step - loss: 0.4548 - acc: 0.7804 - val_loss: 0.6430 - val_acc: 0.6422\n",
            "Epoch 6/60\n",
            "8606/8606 [==============================] - 2s 277us/step - loss: 0.3954 - acc: 0.8147 - val_loss: 0.6536 - val_acc: 0.6284\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f9455516cc0>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gmoNvZS8IaRc",
        "colab_type": "code",
        "outputId": "4127d760-f417-4ff2-aff9-7a43b75e8de1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "cnn.evaluate(test_data, test_labels, batch_size=64, verbose=1)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "860/860 [==============================] - 0s 137us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6114532922589502, 0.6627906973971877]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AHsd1KSzQTL6",
        "colab_type": "code",
        "outputId": "67a0ed69-480d-4634-9659-d0be3cf94a0d",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 484
        }
      },
      "source": [
        "do_analysis(test_data, test_labels, cnn)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.78      0.74      0.76       620\n",
            "           1       0.41      0.47      0.44       240\n",
            "\n",
            "    accuracy                           0.66       860\n",
            "   macro avg       0.60      0.60      0.60       860\n",
            "weighted avg       0.68      0.66      0.67       860\n",
            "\n",
            "[[0.73709677 0.26290323]\n",
            " [0.52916667 0.47083333]]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAU0AAAEYCAYAAADYs6SAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQdPMrAAHTTOzAhw0zcwK+P+dVkbJGvfZQQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BZgHaH2dZXqg",
        "colab_type": "code",
        "outputId": "567275ed-a90e-467d-e8bc-4d43ec855d65",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 561
        }
      },
      "source": [
        "cnn1 = Sequential()\n",
        "cnn1.add(Embedding(vocab_length, 30, input_length=longest_seq))\n",
        "cnn1.add(Reshape((30,longest_seq,1)))\n",
        "cnn1.add(Convolution2D(32,(1,5),activation=\"relu\"))\n",
        "cnn1.add(MaxPooling2D(pool_size=(2,2)))\n",
        "cnn1.add(Dropout(0.5))\n",
        "cnn1.add(Convolution2D(16,(2,3),activation=\"relu\"))\n",
        "cnn1.add(Dropout(0.5))\n",
        "cnn1.add(MaxPooling2D(pool_size=(2,2)))\n",
        "cnn1.add(Convolution2D(16,(2,2),activation=\"relu\"))\n",
        "cnn1.add(Dropout(0.6))\n",
        "cnn1.add(Flatten())\n",
        "cnn1.add(Dense(train_labels.shape[1],activation='sigmoid'))\n",
        "cnn1.compile(loss='categorical_crossentropy', optimizer='adam',metrics=['accuracy'])\n",
        "cnn1.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_14\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "embedding_14 (Embedding)     (None, 44, 30)            472380    \n",
            "_________________________________________________________________\n",
            "reshape_3 (Reshape)          (None, 30, 44, 1)         0         \n",
            "_________________________________________________________________\n",
            "conv2d_7 (Conv2D)            (None, 30, 40, 32)        192       \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 15, 20, 32)        0         \n",
            "_________________________________________________________________\n",
            "dropout_22 (Dropout)         (None, 15, 20, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 14, 18, 16)        3088      \n",
            "_________________________________________________________________\n",
            "dropout_23 (Dropout)         (None, 14, 18, 16)        0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_4 (MaxPooling2 (None, 7, 9, 16)          0         \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 6, 8, 16)          1040      \n",
            "_________________________________________________________________\n",
            "dropout_24 (Dropout)         (None, 6, 8, 16)          0         \n",
            "_________________________________________________________________\n",
            "flatten_3 (Flatten)          (None, 768)               0         \n",
            "_________________________________________________________________\n",
            "dense_23 (Dense)             (None, 2)                 1538      \n",
            "=================================================================\n",
            "Total params: 478,238\n",
            "Trainable params: 478,238\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TjR0XszWjocv",
        "colab_type": "code",
        "outputId": "7bb38178-2725-4ad9-eaef-e483f228c7da",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 255
        }
      },
      "source": [
        "early_stopping = EarlyStopping(monitor='val_loss', patience=2)\n",
        "cnn1.fit(train_data, train_labels, batch_size=64,epochs=60,verbose=1,validation_data=(val_data,val_labels),callbacks=[early_stopping])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 8606 samples, validate on 4634 samples\n",
            "Epoch 1/60\n",
            "8606/8606 [==============================] - 6s 684us/step - loss: 0.6409 - acc: 0.6633 - val_loss: 0.6403 - val_acc: 0.6677\n",
            "Epoch 2/60\n",
            "8606/8606 [==============================] - 2s 213us/step - loss: 0.6373 - acc: 0.6677 - val_loss: 0.6438 - val_acc: 0.6677\n",
            "Epoch 3/60\n",
            "8606/8606 [==============================] - 2s 213us/step - loss: 0.6315 - acc: 0.6677 - val_loss: 0.6358 - val_acc: 0.6677\n",
            "Epoch 4/60\n",
            "8606/8606 [==============================] - 2s 215us/step - loss: 0.5576 - acc: 0.7139 - val_loss: 0.6057 - val_acc: 0.6770\n",
            "Epoch 5/60\n",
            "8606/8606 [==============================] - 2s 215us/step - loss: 0.4215 - acc: 0.8023 - val_loss: 0.6179 - val_acc: 0.6927\n",
            "Epoch 6/60\n",
            "8606/8606 [==============================] - 2s 214us/step - loss: 0.3070 - acc: 0.8677 - val_loss: 0.6711 - val_acc: 0.6577\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f93eb297780>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 58
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U5D-bMGQjtir",
        "colab_type": "code",
        "outputId": "88c9d214-f210-48c1-c907-68f4ce5155de",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        }
      },
      "source": [
        "cnn1.evaluate(test_data, test_labels, batch_size=64, verbose=1)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "860/860 [==============================] - 0s 91us/step\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.6208050087440845, 0.6941860470660898]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 59
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vVZuKs5UkAj6",
        "colab_type": "code",
        "outputId": "1e517168-5e69-4261-a715-fb1bba446845",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 484
        }
      },
      "source": [
        "do_analysis(test_data, test_labels, cnn)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.78      0.74      0.76       620\n",
            "           1       0.41      0.47      0.44       240\n",
            "\n",
            "    accuracy                           0.66       860\n",
            "   macro avg       0.60      0.60      0.60       860\n",
            "weighted avg       0.68      0.66      0.67       860\n",
            "\n",
            "[[0.73709677 0.26290323]\n",
            " [0.52916667 0.47083333]]\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAU0AAAEYCAYAAADYs6SAAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQdPMrAAHTTOzAhw0zcwK+P+dVkbJGvfZQQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XN-f7wGBkCaW",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}