tf_siamese.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "b1ade922-da4f-4107-9467-95a1a972a694",
   "metadata": {},
   "source": [
    "## https://www.kaggle.com/sauravjoshi23/text-classification-using-siamesenet-glove\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "1a426512",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import random\n",
    "import warnings\n",
    "import time\n",
    "import datetime\n",
    "import re\n",
    "import string\n",
    "import itertools\n",
    "import pickle\n",
    "import joblib\n",
    "import nltk\n",
    "import csv\n",
    "from nltk.corpus import stopwords, wordnet\n",
    "stop = set(stopwords.words('english'))\n",
    "from wordcloud import WordCloud, STOPWORDS\n",
    "from nltk.stem import WordNetLemmatizer\n",
    "from nltk.tokenize import word_tokenize\n",
    "from nltk.probability import FreqDist\n",
    "from collections import Counter, defaultdict\n",
    "from keras.utils import np_utils\n",
    "\n",
    "import keras.backend as K\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from sklearn.metrics import classification_report\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import roc_auc_score\n",
    "\n",
    "from tensorflow.keras.preprocessing.sequence import pad_sequences\n",
    "from tensorflow.keras.preprocessing.text import Tokenizer\n",
    "from tensorflow.keras.regularizers import l2\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.optimizers import Adam\n",
    "from tensorflow.keras.layers import Conv2D, ZeroPadding2D, Activation, Input, concatenate,Embedding, LSTM\n",
    "from tensorflow.keras.models import Model\n",
    "from tensorflow.keras.layers import BatchNormalization\n",
    "from tensorflow.keras.layers import MaxPooling2D\n",
    "from tensorflow.keras.layers import Concatenate\n",
    "from tensorflow.keras.layers import Lambda, Flatten, Dense\n",
    "from tensorflow.keras.initializers import glorot_uniform\n",
    "from tensorflow.keras.layers import Input, Dense, Flatten, GlobalMaxPool2D, GlobalAvgPool2D, Concatenate, Multiply, Dropout, Subtract, Add, Conv2D\n",
    "\n",
    "from sklearn.manifold import TSNE\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "%matplotlib inline\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "260304ff",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_data_file_m = '/data/parush/stance_mohammed/new_train.txt'\n",
    "test_data_file_m = '/data/parush/stance_mohammed/new_test.txt'\n",
    "TARGETS_m = [ 'Atheism','Climate Change is a Real Concern', 'Feminist Movement','Hillary Clinton', 'Legalization of Abortion', 'Donald Trump']\n",
    "pos_target = 'Donald Trump' #1 label\n",
    "neg_target = 'Atheism'      # 0 label\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "fafa254a",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "585c2fb8",
   "metadata": {},
   "outputs": [],
   "source": [
    "def df_gen_mohammed(train_data_file_m,pos_target,pos_target_label,neg_target,neg_target_label ):\n",
    "    print('The model is learning {} vs {} '.format(pos_target,neg_target))\n",
    "    df = pd.DataFrame(columns=['text', 'target'])\n",
    "    with open(train_data_file_m, 'r') as trainfile:\n",
    "        for line in trainfile:\n",
    "            line = line.replace('#SemST', '').strip()\n",
    "            line = line.split('\\t')\n",
    "            if line[0].strip() != 'ID' and pos_target in line[1].strip():\n",
    "                tweet = line[2]\n",
    "                df.loc[len(df)] = [tweet,pos_target_label]\n",
    "            elif line[0].strip() != 'ID' and neg_target in line[1].strip():\n",
    "                tweet = line[2]\n",
    "                df.loc[len(df)] = [tweet,neg_target_label]\n",
    "    df = df.sample(frac = 1)\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "538dfb8b-f9c6-4f51-b87e-95b42146bb76",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "8a3afee1-9ac4-476e-8696-54552a7f5ea9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The model is learning Donald Trump vs Atheism \n"
     ]
    }
   ],
   "source": [
    "df = df_gen_mohammed(train_data_file_m,pos_target,1,neg_target,0 )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "id": "b2a4d500",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Here is the class distribution \n",
      " 1    530\n",
      "0    513\n",
      "Name: target, dtype: int64\n"
     ]
    }
   ],
   "source": [
    "print(\"Here is the class distribution \\n\",df['target'].value_counts())"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "38dfe75d",
   "metadata": {},
   "source": [
    "# Data Cleaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "id": "fe37df96",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cleaning Data\n",
      "Cleaning Done\n"
     ]
    }
   ],
   "source": [
    "def remove_url(text):\n",
    "    url = re.compile(r'https?://\\S+|www\\.\\S+')\n",
    "    return url.sub(r'', text)\n",
    "\n",
    "def remove_emoji(text):\n",
    "    emoji_pattern = re.compile(\n",
    "        '['\n",
    "        u'\\U0001F600-\\U0001F64F'  # emoticons\n",
    "        u'\\U0001F300-\\U0001F5FF'  # symbols & pictographs\n",
    "        u'\\U0001F680-\\U0001F6FF'  # transport & map symbols\n",
    "        u'\\U0001F1E0-\\U0001F1FF'  # flags (iOS)\n",
    "        u'\\U00002702-\\U000027B0'\n",
    "        u'\\U000024C2-\\U0001F251'\n",
    "        ']+',\n",
    "        flags=re.UNICODE)\n",
    "    return emoji_pattern.sub(r'', text)\n",
    "\n",
    "def remove_html(text):\n",
    "    html = re.compile(r'<.*?>|&([a-z0-9]+|#[0-9]{1,6}|#x[0-9a-f]{1,6});')\n",
    "    return re.sub(html, '', text)\n",
    "\n",
    "def remove_punct(text):\n",
    "    table = str.maketrans('', '', string.punctuation)\n",
    "    return text.translate(table)\n",
    "\n",
    "df['clean_text'] = df['text'].apply(lambda x: remove_url(str(x)))\n",
    "df['clean_text'] = df['clean_text'].apply(lambda x: remove_emoji(str(x)))\n",
    "df['clean_text'] = df['clean_text'].apply(lambda x: remove_html(str(x)))\n",
    "df['clean_text'] = df['clean_text'].apply(lambda x: remove_punct(str(x)))\n",
    "df['clean_text'] = df['clean_text'].apply(word_tokenize)\n",
    "df['clean_text'] = df['clean_text'].apply(lambda x: [word.lower() for word in x])\n",
    "df['clean_text'] = df['clean_text'].apply(lambda x: [word for word in x if word not in stop])\n",
    "df['clean_text'] = df['clean_text'].apply(nltk.tag.pos_tag)\n",
    "\n",
    "def get_wordnet_pos(tag):\n",
    "    if tag.startswith('J'):\n",
    "        return wordnet.ADJ\n",
    "    elif tag.startswith('V'):\n",
    "        return wordnet.VERB\n",
    "    elif tag.startswith('N'):\n",
    "        return wordnet.NOUN\n",
    "    elif tag.startswith('R'):\n",
    "        return wordnet.ADV\n",
    "    else:\n",
    "        return wordnet.NOUN\n",
    "print(\"Cleaning Data\")    \n",
    "df['clean_text'] = df['clean_text'].apply(\n",
    "    lambda x: [(word, get_wordnet_pos(pos_tag)) for (word, pos_tag) in x])\n",
    "wnl = WordNetLemmatizer()\n",
    "df['clean_text'] = df['clean_text'].apply(\n",
    "    lambda x: [wnl.lemmatize(word, tag) for word, tag in x])\n",
    "df['clean_text'] = df['clean_text'].apply(\n",
    "    lambda x: [word for word in x if word not in stop])\n",
    "df['clean_text'] = [' '.join(map(str, l)) for l in df['clean_text']]\n",
    "print(\"Cleaning Done\")"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "e90a6858",
   "metadata": {},
   "source": [
    "# Triple Loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "4ef15558",
   "metadata": {},
   "outputs": [],
   "source": [
    "positive_initial = list(df[df['target'] == 1]['clean_text'])\n",
    "negative_initial = list(df[df['target'] == 0]['clean_text'])\n",
    "\n",
    "# Taking 1000 samples from the entire data\n",
    "pos = positive_initial[:]\n",
    "neg = negative_initial[:]\n",
    "\n",
    "# Creating pairs of data for siamese training => labels for identity loss and class for disaster or non disaster\n",
    "df2 = pd.DataFrame(columns=['Anchor', 'Positive', 'Negative', 'label', 'class'])\n",
    "\n",
    "for data in pos:\n",
    "    a = data\n",
    "    p = random.choice(pos)\n",
    "    n = random.choice(neg)\n",
    "    df2.loc[len(df2)] = [a, p, n, 1, 1]\n",
    "\n",
    "\n",
    "for data in neg:\n",
    "    a = data\n",
    "    p = random.choice(neg)\n",
    "    n = random.choice(pos)\n",
    "    df2.loc[len(df2)] = [a, p, n, 1, 0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "b8acea3a",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The new DF with anchor, positive, negative, label, class is created. The shape is :  (1043, 5)\n"
     ]
    }
   ],
   "source": [
    "print(\"The new DF with anchor, positive, negative, label, class is created. The shape is : \", df2.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5006ce55",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "ba4f373c",
   "metadata": {},
   "outputs": [],
   "source": [
    "X, X_test, y, y_test = train_test_split(df2[['Anchor', 'Positive', 'Negative']], df2[['label', 'class']], test_size=0.2, random_state=0)\n",
    "X_train, X_val, y_train, y_val = train_test_split(X[['Anchor', 'Positive', 'Negative']], y[['label', 'class']], test_size=0.2, random_state=0)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "24410b4d",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The X_train = (667, 3), X_val = (167, 3), X_test = (209, 3), y_train = (667, 2), y_val = (167, 2), y_test = (209, 2)\n"
     ]
    }
   ],
   "source": [
    "print(\"The X_train = {}, X_val = {}, X_test = {}, y_train = {}, y_val = {}, y_test = {}\".format(X_train.shape, X_val.shape, X_test.shape, y_train.shape, y_val.shape, y_test.shape))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "0497d9ac",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "b0b8fe05",
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train['text'] = X_train[['Anchor', 'Positive', 'Negative']].apply(lambda x: str(x[0])+\" \"+str(x[1])+\" \"+str(x[2]), axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "2b3eadfd",
   "metadata": {},
   "source": [
    "# Tokenizer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "4f1f8ee3",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Starting to Tokenize and Padding\n"
     ]
    }
   ],
   "source": [
    "print(\"Starting to Tokenize and Padding\")\n",
    "t = Tokenizer()\n",
    "t.fit_on_texts(X_train['text'].values)\n",
    "\n",
    "X_train['Anchor'] = X_train['Anchor'].astype(str)\n",
    "X_train['Positive'] = X_train['Positive'].astype(str)\n",
    "X_train['Negative'] = X_train['Negative'].astype(str)\n",
    "X_val['Anchor'] = X_val['Anchor'].astype(str)\n",
    "X_val['Positive'] = X_val['Positive'].astype(str)\n",
    "X_val['Negative'] = X_val['Negative'].astype(str)\n",
    "X_test['Anchor'] = X_test['Anchor'].astype(str)\n",
    "X_test['Positive'] = X_test['Positive'].astype(str)\n",
    "X_test['Negative'] = X_test['Negative'].astype(str)\n",
    "\n",
    "train_q1_seq = t.texts_to_sequences(X_train['Anchor'].values)\n",
    "train_q2_seq = t.texts_to_sequences(X_train['Positive'].values)\n",
    "train_q3_seq = t.texts_to_sequences(X_train['Negative'].values)\n",
    "val_q1_seq = t.texts_to_sequences(X_val['Anchor'].values)\n",
    "val_q2_seq = t.texts_to_sequences(X_val['Positive'].values)\n",
    "val_q3_seq = t.texts_to_sequences(X_val['Negative'].values)\n",
    "test_q1_seq = t.texts_to_sequences(X_test['Anchor'].values)\n",
    "test_q2_seq = t.texts_to_sequences(X_test['Positive'].values)\n",
    "test_q3_seq = t.texts_to_sequences(X_test['Negative'].values)\n",
    "\n",
    "max_len = 200\n",
    "train_q1_seq = pad_sequences(train_q1_seq, maxlen=max_len, padding='post')\n",
    "train_q2_seq = pad_sequences(train_q2_seq, maxlen=max_len, padding='post')\n",
    "train_q3_seq = pad_sequences(train_q3_seq, maxlen=max_len, padding='post')\n",
    "val_q1_seq = pad_sequences(val_q1_seq, maxlen=max_len, padding='post')\n",
    "val_q2_seq = pad_sequences(val_q2_seq, maxlen=max_len, padding='post')\n",
    "val_q3_seq = pad_sequences(val_q3_seq, maxlen=max_len, padding='post')\n",
    "test_q1_seq = pad_sequences(test_q1_seq, maxlen=max_len, padding='post')\n",
    "test_q2_seq = pad_sequences(test_q2_seq, maxlen=max_len, padding='post')\n",
    "test_q3_seq = pad_sequences(test_q3_seq, maxlen=max_len, padding='post')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "f72b7097",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading Embedding and making embedding_matrix\n",
      "Found 400000 word vectors.\n"
     ]
    }
   ],
   "source": [
    "print(\"Loading Embedding and making embedding_matrix\")\n",
    "embeddings_index = {}\n",
    "f = open('/data/parush/embeddings/wikipedia/glove.6B.300d.txt')\n",
    "for line in f:\n",
    "    values = line.split()\n",
    "    word = values[0]\n",
    "    coefs = np.asarray(values[1:], dtype='float32')\n",
    "    embeddings_index[word] = coefs\n",
    "f.close()\n",
    "\n",
    "print('Found %s word vectors.' % len(embeddings_index))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "713e99c5-5d28-43f7-9a44-fa3460c3bb73",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "d3263e8d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 400000 word vectors.\n"
     ]
    }
   ],
   "source": [
    "not_present_list = []\n",
    "vocab_size = len(t.word_index) + 1\n",
    "print('Loaded %s word vectors.' % len(embeddings_index))\n",
    "embedding_matrix = np.zeros((vocab_size, len(embeddings_index['no'])))\n",
    "for word, i in t.word_index.items():\n",
    "    if word in embeddings_index.keys():\n",
    "        \n",
    "        embedding_vector = embeddings_index.get(word)\n",
    "    else:\n",
    "        embedding_vector = None\n",
    "        not_present_list.append(word)\n",
    "    if embedding_vector is not None:\n",
    "        embedding_matrix[i] = embedding_vector\n",
    "    else:\n",
    "        embedding_matrix[i] = np.zeros(300)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "3a0555a9",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Embedding matrix shape (3483, 300)\n"
     ]
    }
   ],
   "source": [
    "print(\"Embedding matrix shape\",embedding_matrix.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f9d8355f",
   "metadata": {},
   "source": [
    "# Siamese Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "d248dcca",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Creating Model and Loss Function\n"
     ]
    }
   ],
   "source": [
    "print(\"Creating Model and Loss Function\")\n",
    "def identity_loss(y_true, y_pred):\n",
    "    return K.mean(y_pred)\n",
    "\n",
    "def triplet_loss(x, alpha = 0.2):\n",
    "    # Triplet Loss function.\n",
    "    anchor,positive,negative = x\n",
    "    # distance between the anchor and the positive\n",
    "    pos_dist = K.sum(K.square(anchor-positive),axis=1)\n",
    "    # distance between the anchor and the negative\n",
    "    neg_dist = K.sum(K.square(anchor-negative),axis=1)\n",
    "    # compute loss\n",
    "    basic_loss = pos_dist-neg_dist+alpha\n",
    "    loss = K.maximum(basic_loss,0.0)\n",
    "    return loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "293fe586",
   "metadata": {},
   "outputs": [],
   "source": [
    "def embedding_model():\n",
    "    network = Sequential()\n",
    "    network.add(Embedding(name=\"synopsis_embedd\",input_dim =len(t.word_index)+1, \n",
    "                       output_dim=len(embeddings_index['no']),weights=[embedding_matrix], \n",
    "                       input_length=train_q1_seq.shape[1],trainable=False))\n",
    "    network.add(LSTM(64,return_sequences=True, activation=\"relu\",dropout=0.3))\n",
    "    network.add(Flatten())\n",
    "    network.add(Dense(128, activation='relu',\n",
    "                  kernel_regularizer=l2(1e-3),\n",
    "                  kernel_initializer='he_uniform'))\n",
    "    network.add(Dropout(0.5))\n",
    "    network.add(Dense(2, activation=None,\n",
    "                  kernel_regularizer=l2(1e-3),\n",
    "                  kernel_initializer='he_uniform'))\n",
    "    #Force the encoding to live on the d-dimentional hypershpere\n",
    "    # network.add(Lambda(lambda x: K.l2_normalize(x,axis=-1)))\n",
    "    return network\n",
    "\n",
    "\n",
    "def build_network(base_model):\n",
    "    input_1 = Input(shape=(train_q1_seq.shape[1],))\n",
    "    input_2 = Input(shape=(train_q2_seq.shape[1],))\n",
    "    input_3 = Input(shape=(train_q3_seq.shape[1],))\n",
    "    \n",
    "    A = base_model(input_1)\n",
    "    P = base_model(input_2)\n",
    "    N = base_model(input_3)\n",
    "\n",
    "    loss = Lambda(triplet_loss)([A, P, N])\n",
    "    model = Model(inputs = [input_1, input_2, input_3], outputs = loss)\n",
    "    model.compile(loss = identity_loss, optimizer = Adam(0.001))\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "c56b2f01",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:Layer lstm_1 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.\n",
      "Model: \"model_1\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_4 (InputLayer)            [(None, 200)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_5 (InputLayer)            [(None, 200)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "input_6 (InputLayer)            [(None, 200)]        0                                            \n",
      "__________________________________________________________________________________________________\n",
      "sequential_1 (Sequential)       (None, 2)            2777126     input_4[0][0]                    \n",
      "                                                                 input_5[0][0]                    \n",
      "                                                                 input_6[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "lambda_1 (Lambda)               (None,)              0           sequential_1[0][0]               \n",
      "                                                                 sequential_1[1][0]               \n",
      "                                                                 sequential_1[2][0]               \n",
      "==================================================================================================\n",
      "Total params: 2,777,126\n",
      "Trainable params: 1,732,226\n",
      "Non-trainable params: 1,044,900\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "base_model = embedding_model()\n",
    "model = build_network(base_model)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "9988ba7f",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train_label = np.asarray(y_train['label']).astype('float32')\n",
    "y_val_label = np.asarray(y_val['label']).astype('float32')\n",
    "y_test_label = np.asarray(y_test['label']).astype('float32')\n",
    "\n",
    "y_train_class = np.asarray(y_train['class']).astype('float32')\n",
    "y_val_class = np.asarray(y_val['class']).astype('float32')\n",
    "y_test_class = np.asarray(y_test['class']).astype('float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "ffe45e64",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning from training data and validating from validation data\n",
      "Epoch 1/50\n",
      "11/11 [==============================] - 10s 630ms/step - loss: 0.3714 - val_loss: 0.2669\n",
      "Epoch 2/50\n",
      "11/11 [==============================] - 6s 573ms/step - loss: 0.2307 - val_loss: 0.1888\n",
      "Epoch 3/50\n",
      "11/11 [==============================] - 6s 594ms/step - loss: 0.1763 - val_loss: 0.1553\n",
      "Epoch 4/50\n",
      "11/11 [==============================] - 6s 598ms/step - loss: 0.1489 - val_loss: 0.1482\n",
      "Epoch 5/50\n",
      "11/11 [==============================] - 6s 577ms/step - loss: 0.1238 - val_loss: 0.1312\n",
      "Epoch 6/50\n",
      "11/11 [==============================] - 6s 589ms/step - loss: 0.1057 - val_loss: 0.1217\n",
      "Epoch 7/50\n",
      "11/11 [==============================] - 6s 562ms/step - loss: 0.0959 - val_loss: 0.1207\n",
      "Epoch 8/50\n",
      "11/11 [==============================] - 6s 590ms/step - loss: 0.0781 - val_loss: 0.1174\n",
      "Epoch 9/50\n",
      "11/11 [==============================] - 6s 596ms/step - loss: 0.0719 - val_loss: 0.1165\n",
      "Epoch 10/50\n",
      "11/11 [==============================] - 6s 591ms/step - loss: 0.0584 - val_loss: 0.1146\n",
      "Epoch 11/50\n",
      "11/11 [==============================] - 6s 564ms/step - loss: 0.0559 - val_loss: 0.1168\n",
      "Epoch 12/50\n",
      "11/11 [==============================] - 6s 591ms/step - loss: 0.0517 - val_loss: 0.1084\n",
      "Epoch 13/50\n",
      "11/11 [==============================] - 6s 586ms/step - loss: 0.0489 - val_loss: 0.1042\n",
      "Epoch 14/50\n",
      "11/11 [==============================] - 6s 577ms/step - loss: 0.0476 - val_loss: 0.1024\n",
      "Epoch 15/50\n",
      "11/11 [==============================] - 6s 583ms/step - loss: 0.0398 - val_loss: 0.1010\n",
      "Epoch 16/50\n",
      "11/11 [==============================] - 6s 574ms/step - loss: 0.0402 - val_loss: 0.1056\n",
      "Epoch 17/50\n",
      "11/11 [==============================] - 6s 569ms/step - loss: 0.0471 - val_loss: 0.0928\n",
      "Epoch 18/50\n",
      "11/11 [==============================] - 6s 575ms/step - loss: 0.0359 - val_loss: 0.0934\n",
      "Epoch 19/50\n",
      "11/11 [==============================] - 6s 590ms/step - loss: 0.0320 - val_loss: 0.1063\n",
      "Epoch 20/50\n",
      "11/11 [==============================] - 6s 553ms/step - loss: 0.0334 - val_loss: 0.1099\n",
      "Epoch 21/50\n",
      "11/11 [==============================] - 6s 581ms/step - loss: 0.0262 - val_loss: 0.1074\n",
      "Epoch 22/50\n",
      "11/11 [==============================] - 6s 580ms/step - loss: 0.0338 - val_loss: 0.1082\n",
      "Epoch 23/50\n",
      "11/11 [==============================] - 6s 584ms/step - loss: 0.0267 - val_loss: 0.1184\n",
      "Epoch 24/50\n",
      "11/11 [==============================] - 6s 571ms/step - loss: 0.0223 - val_loss: 0.1226\n",
      "Epoch 25/50\n",
      "11/11 [==============================] - 6s 564ms/step - loss: 0.0258 - val_loss: 0.1212\n",
      "Epoch 26/50\n",
      "11/11 [==============================] - 7s 611ms/step - loss: 0.0332 - val_loss: 0.1231\n",
      "Epoch 27/50\n",
      "11/11 [==============================] - 6s 553ms/step - loss: 0.0250 - val_loss: 0.1372\n",
      "Epoch 28/50\n",
      "11/11 [==============================] - 6s 589ms/step - loss: 0.0267 - val_loss: 0.1456\n",
      "Epoch 29/50\n",
      "11/11 [==============================] - 6s 578ms/step - loss: 0.0341 - val_loss: 0.1258\n",
      "Epoch 30/50\n",
      "11/11 [==============================] - 6s 597ms/step - loss: 0.0278 - val_loss: 0.1196\n",
      "Epoch 31/50\n",
      "11/11 [==============================] - 6s 590ms/step - loss: 0.0297 - val_loss: 0.1141\n",
      "Epoch 32/50\n",
      "11/11 [==============================] - 6s 580ms/step - loss: 0.0239 - val_loss: 0.1174\n",
      "Epoch 33/50\n",
      "11/11 [==============================] - 7s 610ms/step - loss: 0.0280 - val_loss: 0.1219\n",
      "Epoch 34/50\n",
      "11/11 [==============================] - 6s 597ms/step - loss: 0.0232 - val_loss: 0.1292\n",
      "Epoch 35/50\n",
      "11/11 [==============================] - 6s 585ms/step - loss: 0.0223 - val_loss: 0.1220\n",
      "Epoch 36/50\n",
      "11/11 [==============================] - 7s 607ms/step - loss: 0.0235 - val_loss: 0.1109\n",
      "Epoch 37/50\n",
      "11/11 [==============================] - 7s 592ms/step - loss: 0.0253 - val_loss: 0.1071\n",
      "Epoch 38/50\n",
      "11/11 [==============================] - 7s 602ms/step - loss: 0.0219 - val_loss: 0.1097\n",
      "Epoch 39/50\n",
      "11/11 [==============================] - 7s 604ms/step - loss: 0.0204 - val_loss: 0.1209\n",
      "Epoch 40/50\n",
      "11/11 [==============================] - 7s 600ms/step - loss: 0.0244 - val_loss: 0.1135\n",
      "Epoch 41/50\n",
      "11/11 [==============================] - 7s 607ms/step - loss: 0.0179 - val_loss: 0.1107\n",
      "Epoch 42/50\n",
      "11/11 [==============================] - 6s 589ms/step - loss: 0.0184 - val_loss: 0.1204\n",
      "Epoch 43/50\n",
      "11/11 [==============================] - 6s 596ms/step - loss: 0.0231 - val_loss: 0.1229\n",
      "Epoch 44/50\n",
      "11/11 [==============================] - 6s 585ms/step - loss: 0.0206 - val_loss: 0.1193\n",
      "Epoch 45/50\n",
      "11/11 [==============================] - 7s 593ms/step - loss: 0.0177 - val_loss: 0.1209\n",
      "Epoch 46/50\n",
      "11/11 [==============================] - 6s 596ms/step - loss: 0.0159 - val_loss: 0.1267\n",
      "Epoch 47/50\n",
      "11/11 [==============================] - 6s 596ms/step - loss: 0.0180 - val_loss: 0.1330\n",
      "Epoch 48/50\n",
      "11/11 [==============================] - 7s 594ms/step - loss: 0.0203 - val_loss: 0.1471\n",
      "Epoch 49/50\n",
      "11/11 [==============================] - 6s 582ms/step - loss: 0.0188 - val_loss: 0.1574\n",
      "Epoch 50/50\n",
      "11/11 [==============================] - 7s 597ms/step - loss: 0.0177 - val_loss: 0.1542\n"
     ]
    }
   ],
   "source": [
    "print('Learning from training data and validating from validation data')\n",
    "history = model.fit([train_q1_seq,train_q2_seq, train_q3_seq],y_train_label.reshape(-1,1), epochs = 50, \n",
    "          batch_size=64,validation_data=([val_q1_seq, val_q2_seq, val_q3_seq],y_val_label.reshape(-1,1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "8f587744",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('Training and Validation Losses',size = 12)\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'val'], loc='upper right')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "9926a334",
   "metadata": {},
   "outputs": [],
   "source": [
    "# model_json = model.to_json()\n",
    "# with open(\"siamesemodel-triplet-loss.json\", \"w\") as json_file:\n",
    "#     json_file.write(model_json)\n",
    "# #serialize weights to HDF5\n",
    "# model.save_weights(\"siamesemodel-triplet-loss.h5\")\n",
    "# print(\"Saved model to disk\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "50ed38fe",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "X_train_eval = base_model.predict(train_q1_seq)\n",
    "X_test_eval = base_model.predict(test_q1_seq)\n",
    "# The above embedding outputs will be used to predict. \n",
    "\n",
    "# TSNE - dimensionality reduction for data visualization\n",
    "tsne = TSNE()\n",
    "train_tsne_embeds = tsne.fit_transform(X_train_eval)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "754a9577",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "def scatter(x, labels, subtitle=None):\n",
    "    # Create a scatter plot of all the \n",
    "    # the embeddings of the model.\n",
    "    # We choose a color palette with seaborn.\n",
    "    palette = np.array(sns.color_palette(\"hls\", 2))\n",
    "    # We create a scatter plot.\n",
    "    f = plt.figure(figsize=(8, 8))\n",
    "    ax = plt.subplot(aspect='equal')\n",
    "    sc = ax.scatter(x[:,0], x[:,1], lw=0,alpha = 0.5, s=40,\n",
    "                    c=palette[labels.astype(np.int)] )\n",
    "    plt.xlim(-25, 25)\n",
    "    plt.ylim(-25, 25)\n",
    "    ax.axis('off')\n",
    "    ax.axis('tight')\n",
    "\n",
    "scatter(train_tsne_embeds, y_train_class)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "5bdd7564-88f1-44b6-8aec-b2f7c2c8bf2f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "id": "39efcd6c",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "21/21 [==============================] - 0s 7ms/step - loss: 0.3568 - accuracy: 0.9880 - val_loss: 0.4742 - val_accuracy: 0.8804\n",
      "Epoch 2/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3475 - accuracy: 0.9880 - val_loss: 0.4699 - val_accuracy: 0.8804\n",
      "Epoch 3/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3387 - accuracy: 0.9895 - val_loss: 0.4658 - val_accuracy: 0.8852\n",
      "Epoch 4/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3305 - accuracy: 0.9910 - val_loss: 0.4619 - val_accuracy: 0.8852\n",
      "Epoch 5/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3226 - accuracy: 0.9910 - val_loss: 0.4583 - val_accuracy: 0.8804\n",
      "Epoch 6/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3151 - accuracy: 0.9910 - val_loss: 0.4550 - val_accuracy: 0.8852\n",
      "Epoch 7/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3079 - accuracy: 0.9910 - val_loss: 0.4517 - val_accuracy: 0.8756\n",
      "Epoch 8/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.3010 - accuracy: 0.9895 - val_loss: 0.4487 - val_accuracy: 0.8756\n",
      "Epoch 9/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2945 - accuracy: 0.9895 - val_loss: 0.4457 - val_accuracy: 0.8756\n",
      "Epoch 10/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2882 - accuracy: 0.9895 - val_loss: 0.4429 - val_accuracy: 0.8756\n",
      "Epoch 11/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2822 - accuracy: 0.9895 - val_loss: 0.4402 - val_accuracy: 0.8756\n",
      "Epoch 12/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2765 - accuracy: 0.9895 - val_loss: 0.4378 - val_accuracy: 0.8756\n",
      "Epoch 13/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2708 - accuracy: 0.9895 - val_loss: 0.4353 - val_accuracy: 0.8756\n",
      "Epoch 14/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2655 - accuracy: 0.9895 - val_loss: 0.4330 - val_accuracy: 0.8708\n",
      "Epoch 15/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2604 - accuracy: 0.9895 - val_loss: 0.4309 - val_accuracy: 0.8708\n",
      "Epoch 16/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2555 - accuracy: 0.9895 - val_loss: 0.4288 - val_accuracy: 0.8708\n",
      "Epoch 17/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2507 - accuracy: 0.9895 - val_loss: 0.4271 - val_accuracy: 0.8708\n",
      "Epoch 18/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2461 - accuracy: 0.9895 - val_loss: 0.4252 - val_accuracy: 0.8708\n",
      "Epoch 19/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2417 - accuracy: 0.9895 - val_loss: 0.4235 - val_accuracy: 0.8708\n",
      "Epoch 20/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2374 - accuracy: 0.9895 - val_loss: 0.4219 - val_accuracy: 0.8708\n",
      "Epoch 21/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2333 - accuracy: 0.9895 - val_loss: 0.4202 - val_accuracy: 0.8708\n",
      "Epoch 22/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2293 - accuracy: 0.9895 - val_loss: 0.4187 - val_accuracy: 0.8708\n",
      "Epoch 23/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2255 - accuracy: 0.9895 - val_loss: 0.4174 - val_accuracy: 0.8708\n",
      "Epoch 24/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2218 - accuracy: 0.9895 - val_loss: 0.4159 - val_accuracy: 0.8708\n",
      "Epoch 25/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2182 - accuracy: 0.9895 - val_loss: 0.4148 - val_accuracy: 0.8708\n",
      "Epoch 26/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2147 - accuracy: 0.9895 - val_loss: 0.4135 - val_accuracy: 0.8708\n",
      "Epoch 27/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2114 - accuracy: 0.9895 - val_loss: 0.4125 - val_accuracy: 0.8708\n",
      "Epoch 28/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2081 - accuracy: 0.9895 - val_loss: 0.4115 - val_accuracy: 0.8708\n",
      "Epoch 29/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2050 - accuracy: 0.9895 - val_loss: 0.4106 - val_accuracy: 0.8708\n",
      "Epoch 30/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.2019 - accuracy: 0.9895 - val_loss: 0.4096 - val_accuracy: 0.8708\n",
      "Epoch 31/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1990 - accuracy: 0.9910 - val_loss: 0.4087 - val_accuracy: 0.8708\n",
      "Epoch 32/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1961 - accuracy: 0.9910 - val_loss: 0.4078 - val_accuracy: 0.8708\n",
      "Epoch 33/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1934 - accuracy: 0.9910 - val_loss: 0.4071 - val_accuracy: 0.8708\n",
      "Epoch 34/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1907 - accuracy: 0.9910 - val_loss: 0.4063 - val_accuracy: 0.8708\n",
      "Epoch 35/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1881 - accuracy: 0.9910 - val_loss: 0.4057 - val_accuracy: 0.8708\n",
      "Epoch 36/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1855 - accuracy: 0.9910 - val_loss: 0.4052 - val_accuracy: 0.8708\n",
      "Epoch 37/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1831 - accuracy: 0.9910 - val_loss: 0.4045 - val_accuracy: 0.8708\n",
      "Epoch 38/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1807 - accuracy: 0.9910 - val_loss: 0.4040 - val_accuracy: 0.8708\n",
      "Epoch 39/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1784 - accuracy: 0.9910 - val_loss: 0.4034 - val_accuracy: 0.8708\n",
      "Epoch 40/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1762 - accuracy: 0.9910 - val_loss: 0.4030 - val_accuracy: 0.8708\n",
      "Epoch 41/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1740 - accuracy: 0.9910 - val_loss: 0.4026 - val_accuracy: 0.8708\n",
      "Epoch 42/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1718 - accuracy: 0.9910 - val_loss: 0.4023 - val_accuracy: 0.8708\n",
      "Epoch 43/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1698 - accuracy: 0.9910 - val_loss: 0.4019 - val_accuracy: 0.8708\n",
      "Epoch 44/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1678 - accuracy: 0.9910 - val_loss: 0.4017 - val_accuracy: 0.8708\n",
      "Epoch 45/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1658 - accuracy: 0.9910 - val_loss: 0.4014 - val_accuracy: 0.8708\n",
      "Epoch 46/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1640 - accuracy: 0.9910 - val_loss: 0.4011 - val_accuracy: 0.8708\n",
      "Epoch 47/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1621 - accuracy: 0.9910 - val_loss: 0.4009 - val_accuracy: 0.8708\n",
      "Epoch 48/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1603 - accuracy: 0.9910 - val_loss: 0.4008 - val_accuracy: 0.8708\n",
      "Epoch 49/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1586 - accuracy: 0.9910 - val_loss: 0.4005 - val_accuracy: 0.8708\n",
      "Epoch 50/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1569 - accuracy: 0.9910 - val_loss: 0.4004 - val_accuracy: 0.8708\n",
      "Epoch 51/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1552 - accuracy: 0.9910 - val_loss: 0.4002 - val_accuracy: 0.8708\n",
      "Epoch 52/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1536 - accuracy: 0.9910 - val_loss: 0.4002 - val_accuracy: 0.8708\n",
      "Epoch 53/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1520 - accuracy: 0.9910 - val_loss: 0.4001 - val_accuracy: 0.8708\n",
      "Epoch 54/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1505 - accuracy: 0.9910 - val_loss: 0.4001 - val_accuracy: 0.8708\n",
      "Epoch 55/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1490 - accuracy: 0.9910 - val_loss: 0.4001 - val_accuracy: 0.8708\n",
      "Epoch 56/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1475 - accuracy: 0.9910 - val_loss: 0.4001 - val_accuracy: 0.8708\n",
      "Epoch 57/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1461 - accuracy: 0.9910 - val_loss: 0.4000 - val_accuracy: 0.8708\n",
      "Epoch 58/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1447 - accuracy: 0.9910 - val_loss: 0.4002 - val_accuracy: 0.8708\n",
      "Epoch 59/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1434 - accuracy: 0.9910 - val_loss: 0.4002 - val_accuracy: 0.8708\n",
      "Epoch 60/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1421 - accuracy: 0.9910 - val_loss: 0.4002 - val_accuracy: 0.8708\n",
      "Epoch 61/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1408 - accuracy: 0.9910 - val_loss: 0.4004 - val_accuracy: 0.8708\n",
      "Epoch 62/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1395 - accuracy: 0.9910 - val_loss: 0.4005 - val_accuracy: 0.8708\n",
      "Epoch 63/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1383 - accuracy: 0.9910 - val_loss: 0.4005 - val_accuracy: 0.8708\n",
      "Epoch 64/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1371 - accuracy: 0.9910 - val_loss: 0.4007 - val_accuracy: 0.8708\n",
      "Epoch 65/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1359 - accuracy: 0.9910 - val_loss: 0.4009 - val_accuracy: 0.8708\n",
      "Epoch 66/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1348 - accuracy: 0.9910 - val_loss: 0.4011 - val_accuracy: 0.8708\n",
      "Epoch 67/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1337 - accuracy: 0.9910 - val_loss: 0.4012 - val_accuracy: 0.8708\n",
      "Epoch 68/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1326 - accuracy: 0.9910 - val_loss: 0.4014 - val_accuracy: 0.8708\n",
      "Epoch 69/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1316 - accuracy: 0.9910 - val_loss: 0.4016 - val_accuracy: 0.8708\n",
      "Epoch 70/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1305 - accuracy: 0.9910 - val_loss: 0.4018 - val_accuracy: 0.8708\n",
      "Epoch 71/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1295 - accuracy: 0.9910 - val_loss: 0.4021 - val_accuracy: 0.8708\n",
      "Epoch 72/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1285 - accuracy: 0.9910 - val_loss: 0.4023 - val_accuracy: 0.8708\n",
      "Epoch 73/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1275 - accuracy: 0.9910 - val_loss: 0.4026 - val_accuracy: 0.8708\n",
      "Epoch 74/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1266 - accuracy: 0.9910 - val_loss: 0.4029 - val_accuracy: 0.8708\n",
      "Epoch 75/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1256 - accuracy: 0.9910 - val_loss: 0.4031 - val_accuracy: 0.8708\n",
      "Epoch 76/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1248 - accuracy: 0.9910 - val_loss: 0.4035 - val_accuracy: 0.8708\n",
      "Epoch 77/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1239 - accuracy: 0.9910 - val_loss: 0.4037 - val_accuracy: 0.8708\n",
      "Epoch 78/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1230 - accuracy: 0.9910 - val_loss: 0.4041 - val_accuracy: 0.8708\n",
      "Epoch 79/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1221 - accuracy: 0.9910 - val_loss: 0.4043 - val_accuracy: 0.8708\n",
      "Epoch 80/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1213 - accuracy: 0.9910 - val_loss: 0.4046 - val_accuracy: 0.8708\n",
      "Epoch 81/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1205 - accuracy: 0.9910 - val_loss: 0.4050 - val_accuracy: 0.8708\n",
      "Epoch 82/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1197 - accuracy: 0.9910 - val_loss: 0.4053 - val_accuracy: 0.8708\n",
      "Epoch 83/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1189 - accuracy: 0.9910 - val_loss: 0.4057 - val_accuracy: 0.8708\n",
      "Epoch 84/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1182 - accuracy: 0.9910 - val_loss: 0.4060 - val_accuracy: 0.8708\n",
      "Epoch 85/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1174 - accuracy: 0.9910 - val_loss: 0.4064 - val_accuracy: 0.8708\n",
      "Epoch 86/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1166 - accuracy: 0.9910 - val_loss: 0.4068 - val_accuracy: 0.8708\n",
      "Epoch 87/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1159 - accuracy: 0.9910 - val_loss: 0.4072 - val_accuracy: 0.8708\n",
      "Epoch 88/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1153 - accuracy: 0.9910 - val_loss: 0.4076 - val_accuracy: 0.8756\n",
      "Epoch 89/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1146 - accuracy: 0.9910 - val_loss: 0.4080 - val_accuracy: 0.8708\n",
      "Epoch 90/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1139 - accuracy: 0.9910 - val_loss: 0.4084 - val_accuracy: 0.8756\n",
      "Epoch 91/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1132 - accuracy: 0.9910 - val_loss: 0.4087 - val_accuracy: 0.8756\n",
      "Epoch 92/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1126 - accuracy: 0.9910 - val_loss: 0.4092 - val_accuracy: 0.8756\n",
      "Epoch 93/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1119 - accuracy: 0.9910 - val_loss: 0.4096 - val_accuracy: 0.8756\n",
      "Epoch 94/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1113 - accuracy: 0.9910 - val_loss: 0.4100 - val_accuracy: 0.8756\n",
      "Epoch 95/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1107 - accuracy: 0.9910 - val_loss: 0.4104 - val_accuracy: 0.8756\n",
      "Epoch 96/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1101 - accuracy: 0.9910 - val_loss: 0.4108 - val_accuracy: 0.8756\n",
      "Epoch 97/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1095 - accuracy: 0.9910 - val_loss: 0.4112 - val_accuracy: 0.8756\n",
      "Epoch 98/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1090 - accuracy: 0.9910 - val_loss: 0.4118 - val_accuracy: 0.8756\n",
      "Epoch 99/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1084 - accuracy: 0.9910 - val_loss: 0.4121 - val_accuracy: 0.8756\n",
      "Epoch 100/100\n",
      "21/21 [==============================] - 0s 3ms/step - loss: 0.1079 - accuracy: 0.9910 - val_loss: 0.4126 - val_accuracy: 0.8756\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f2438f1d910>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# classifier that computes the class of a specific embedding for prediction purposes\n",
    "classifier_input = Input(shape=(2,))\n",
    "classifier_output = Dense(2, activation='softmax')(classifier_input)\n",
    "classifier_model = Model(classifier_input, classifier_output)\n",
    "\n",
    "# onehot encoding vectors to 2 classes\n",
    "Y_train_onehot = np_utils.to_categorical(y_train_class, 2)\n",
    "Y_test_onehot = np_utils.to_categorical(y_test_class, 2)\n",
    "\n",
    "classifier_model.compile(optimizer='adam',loss='categorical_crossentropy',metrics=['accuracy'])\n",
    "classifier_model.fit(X_train_eval,Y_train_onehot, validation_data=(X_test_eval, Y_test_onehot),epochs=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "aa8dafaf",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Classification report on validation data \n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0     0.8938    0.8783    0.8860       115\n",
      "           1     0.8542    0.8723    0.8632        94\n",
      "\n",
      "    accuracy                         0.8756       209\n",
      "   macro avg     0.8740    0.8753    0.8746       209\n",
      "weighted avg     0.8760    0.8756    0.8757       209\n",
      "\n"
     ]
    }
   ],
   "source": [
    "\n",
    "input = X_test_eval\n",
    "y_pred = classifier_model.predict(input)\n",
    "y_pred=np.argmax(y_pred, axis=1)\n",
    "y_test = np.argmax(Y_test_onehot,axis = 1)\n",
    "print(\"Classification report on validation data \\n\",classification_report(y_test, y_pred, digits=4))\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "id": "2156e31f",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Formating test file for Donald Trump vs Atheism \n",
      "Testing on Donald Trump as label 1 and\n",
      "Testing on Atheism as label 0 and\n"
     ]
    }
   ],
   "source": [
    "print(\"Formating test file for {} vs {} \".format(pos_target,neg_target))\n",
    "df_test = pd.DataFrame(columns=['text', 'label', 'stance'])\n",
    "label = 1\n",
    "for target in [pos_target,neg_target]:\n",
    "    print('Testing on {} as label {} and'.format(target,label))\n",
    "    with open(test_data_file_m, 'r') as testfile:\n",
    "        for line in testfile:\n",
    "            line = line.replace('#SemST', '').strip()\n",
    "            line = line.split('\\t')\n",
    "            if line[0].strip() != 'ID' and target in line[1].strip():\n",
    "                tweet = line[2]\n",
    "                df_test.loc[len(df_test)] = [str(tweet),label, line[3]]\n",
    "    label = 0\n",
    "df_test['clean_text'] = df_test['text'].apply(lambda x: remove_url(str(x)))\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(lambda x: remove_emoji(str(x)))\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(lambda x: remove_html(str(x)))\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(lambda x: remove_punct(str(x)))\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(word_tokenize)\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(lambda x: [word.lower() for word in x])\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(lambda x: [word for word in x if word not in stop])\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(nltk.tag.pos_tag)\n",
    "\n",
    "def get_wordnet_pos(tag):\n",
    "    if tag.startswith('J'):\n",
    "        return wordnet.ADJ\n",
    "    elif tag.startswith('V'):\n",
    "        return wordnet.VERB\n",
    "    elif tag.startswith('N'):\n",
    "        return wordnet.NOUN\n",
    "    elif tag.startswith('R'):\n",
    "        return wordnet.ADV\n",
    "    else:\n",
    "        return wordnet.NOUN\n",
    "    \n",
    "df_test['clean_text'] = df_test['clean_text'].apply(\n",
    "    lambda x: [(word, get_wordnet_pos(pos_tag)) for (word, pos_tag) in x])\n",
    "wnl = WordNetLemmatizer()\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(\n",
    "    lambda x: [wnl.lemmatize(word, tag) for word, tag in x])\n",
    "df_test['clean_text'] = df_test['clean_text'].apply(\n",
    "    lambda x: [word for word in x if word not in stop])\n",
    "df_test['clean_text'] = [' '.join(map(str, l)) for l in df_test['clean_text']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "id": "2614ac42",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Classification report on Test data \n",
      "               precision    recall  f1-score   support\n",
      "\n",
      "           0     0.8957    0.8591    0.8770       220\n",
      "           1     0.8333    0.8757    0.8540       177\n",
      "\n",
      "    accuracy                         0.8665       397\n",
      "   macro avg     0.8645    0.8674    0.8655       397\n",
      "weighted avg     0.8679    0.8665    0.8668       397\n",
      "\n"
     ]
    }
   ],
   "source": [
    "df_test['clean_text'] = df_test['clean_text'].astype(str)\n",
    "y_q1_seq = t.texts_to_sequences(df_test['clean_text'].values)\n",
    "\n",
    "\n",
    "y_q1_seq = pad_sequences(y_q1_seq, maxlen=max_len, padding='post')\n",
    "y_test_eval = base_model.predict(y_q1_seq)\n",
    "\n",
    "input_2 = y_test_eval\n",
    "y_pred_2 = classifier_model.predict(input_2)\n",
    "\n",
    "y2_test_class = np.asarray(df_test['label']).astype('float32')\n",
    "y2_test_onehot = np_utils.to_categorical(y2_test_class, 2)\n",
    "\n",
    "y_pred_2=np.argmax(y_pred_2, axis=1)\n",
    "y_test_2 = np.argmax(y2_test_onehot,axis = 1)\n",
    "print(\"\")\n",
    "print(\"Classification report on Test data \\n\",classification_report(y_test_2, y_pred_2, digits=4))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f352a8f0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b2d63e44",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "1b2e8c43",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "f288c20d-1cac-4160-aa4e-fbb7f8b42c1f",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6bac4379",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d9d31daf-26df-46ea-8bb8-ce70182afbd8",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d02a5278",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ad824d7d",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b13766df",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "68204cb4",
   "metadata": {},
   "source": [
    "# predicting above model on other target and getting that data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "id": "09df8135-d10d-47ea-8dc7-a8bdedcce653",
   "metadata": {},
   "outputs": [],
   "source": [
    "TARGETS_m = [ 'Atheism','Climate Change is a Real Concern', 'Feminist Movement','Hillary Clinton', 'Legalization of Abortion', 'Donald Trump']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "id": "970a9ec7",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_on = 'Hillary Clinton'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "id": "25522757",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_oth = pd.DataFrame(columns=['text', 'label', 'stance'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "id": "4ee49ba6",
   "metadata": {},
   "outputs": [],
   "source": [
    " with open(train_data_file_m, 'r') as testfile:\n",
    "        for line in testfile:\n",
    "            \n",
    "            line = line.replace('#SemST', '').strip()\n",
    "            line = line.split('\\t')\n",
    "            \n",
    "            if line[0].strip() != 'ID' and test_on in line[1].strip():\n",
    "                tweet = line[2]\n",
    "                #print(tweet)\n",
    "                df_oth.loc[len(df_oth)] = [str(tweet),'NA',line[3]]\n",
    "                "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4a65a0e0",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "id": "bef43359",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_oth['clean_text'] = df_oth['text'].apply(lambda x: remove_url(str(x)))\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(lambda x: remove_emoji(str(x)))\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(lambda x: remove_html(str(x)))\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(lambda x: remove_punct(str(x)))\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(word_tokenize)\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(lambda x: [word.lower() for word in x])\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(lambda x: [word for word in x if word not in stop])\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(nltk.tag.pos_tag)\n",
    "\n",
    "def get_wordnet_pos(tag):\n",
    "    if tag.startswith('J'):\n",
    "        return wordnet.ADJ\n",
    "    elif tag.startswith('V'):\n",
    "        return wordnet.VERB\n",
    "    elif tag.startswith('N'):\n",
    "        return wordnet.NOUN\n",
    "    elif tag.startswith('R'):\n",
    "        return wordnet.ADV\n",
    "    else:\n",
    "        return wordnet.NOUN\n",
    "    \n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(\n",
    "    lambda x: [(word, get_wordnet_pos(pos_tag)) for (word, pos_tag) in x])\n",
    "wnl = WordNetLemmatizer()\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(\n",
    "    lambda x: [wnl.lemmatize(word, tag) for word, tag in x])\n",
    "df_oth['clean_text'] = df_oth['clean_text'].apply(\n",
    "    lambda x: [word for word in x if word not in stop])\n",
    "df_oth['clean_text'] = [' '.join(map(str, l)) for l in df_oth['clean_text']]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "id": "a37f8ca7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'AGAINST'"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_oth['stance'][5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "id": "d23d2b5a",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_oth['clean_text'] = df_oth['clean_text'].astype(str)\n",
    "oth_q1_seq = t.texts_to_sequences(df_oth['clean_text'].values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "id": "52b17e7d",
   "metadata": {},
   "outputs": [],
   "source": [
    "oth_q1_seq = pad_sequences(oth_q1_seq, maxlen=max_len, padding='post')\n",
    "oth_test_eval = base_model.predict(oth_q1_seq)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "id": "8513bbe8",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_3 = oth_test_eval\n",
    "oth_pred_3 = classifier_model.predict(input_3)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "id": "91c15ac2-5048-433b-8c88-ff2114bc37b9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.29488093, 0.705119  ],\n",
       "       [0.11479663, 0.8852033 ],\n",
       "       [0.09600843, 0.9039916 ],\n",
       "       ...,\n",
       "       [0.6151274 , 0.38487265],\n",
       "       [0.47174916, 0.5282509 ],\n",
       "       [0.04328242, 0.9567176 ]], dtype=float32)"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "oth_pred_3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "077bf2b9-4bf9-45fb-8acc-3716642e2130",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "302bb6a5-9230-41bb-8f38-d59f1ac15aaf",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "id": "934c5b48",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1,\n",
       "       1, 1, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1,\n",
       "       1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0,\n",
       "       1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1,\n",
       "       1, 0, 1, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1,\n",
       "       1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 0, 1, 1, 1, 0, 1, 0, 1, 1, 1, 1, 0, 1, 1, 0, 1, 0, 0, 0,\n",
       "       1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 0, 0, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 0, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1,\n",
       "       1, 1, 1, 0, 0, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 0,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1, 1,\n",
       "       0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1,\n",
       "       1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1,\n",
       "       0, 1, 0, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1,\n",
       "       1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 0, 1, 1, 1, 1,\n",
       "       1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,\n",
       "       1, 1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 1, 0, 1, 1,\n",
       "       1, 1, 0, 0, 0, 1, 1])"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "oth_pred_4=np.argmax(oth_pred_3, axis=1)\n",
    "oth_pred_4"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "id": "249959d0-cb8a-4010-b92a-aa8f56d93096",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.29488093, 0.705119  ], dtype=float32)"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "oth_pred_3[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "id": "a76e5fe5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0: 97, 1: 592}"
      ]
     },
     "execution_count": 61,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "unique,count = np.unique(oth_pred_4,return_counts=True)\n",
    "dict(zip(unique, count))"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "b7913f01-aed4-4c9e-8f34-9092a4580d4d",
   "metadata": {
    "tags": []
   },
   "source": [
    "# Change the file name "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "94ed4734-8aaa-4225-b6b6-b6126ae52fb0",
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "    \n",
    "        \n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "id": "f86ce53f",
   "metadata": {},
   "outputs": [],
   "source": [
    "th = 0.50\n",
    "df_rich_data= pd.DataFrame(columns = ['Tweet','Stance'])\n",
    "file_name = 'rich_data/dt_ath_hc/' + pos_target + '_' + neg_target + '_' +test_on + str(th) +'.txt'\n",
    "with open(file_name,'w') as f:\n",
    "    for i in range(len(oth_pred_3)):\n",
    "            if oth_pred_4[i] == 1:\n",
    "                if oth_pred_3[i][1] >= th:\n",
    "                    f.write(df_oth['text'][i] + \"\\t\" +df_oth['stance'][i])\n",
    "                    f.write('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "id": "9577c8d9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "92"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "c = 0\n",
    "with open(file_name,'r') as f:\n",
    "    for i in f:\n",
    "        c += 1\n",
    "c"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "id": "10e20262-9020-47a1-8f15-5209bf7522de",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "rich_data/dt_ath_hc/Donald Trump_Atheism_Hillary Clinton0.5.txt\n"
     ]
    }
   ],
   "source": [
    "print(file_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "id": "58a212d2",
   "metadata": {},
   "outputs": [],
   "source": [
    "# with open('rich_data.txt','w') as f:\n",
    "#     for tweet in rich_data:\n",
    "#         f.write(tweet)\n",
    "#         f.write('\\n')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "id": "522621f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_rich_data.to_csv('rich_data.csv',sep='\\t')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "23dc36b7",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "nlpKernel",
   "language": "python",
   "name": "nlpkernel"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}