supervised_3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## BERT embedding and final results "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "I trained a classification NN here after BERT embeddings. BERT, based on transformers, is very computationally intensive. I will use a subset of my dataset for training, this will of course reduce the performance but it is the limit of my computer resources.  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "c:\\dev\\topic_modelling\\lib\\site-packages\\umap\\distances.py:1063: NumbaDeprecationWarning: \u001b[1mThe 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.\u001b[0m\n",
      "  @numba.jit()\n",
      "c:\\dev\\topic_modelling\\lib\\site-packages\\umap\\distances.py:1071: NumbaDeprecationWarning: \u001b[1mThe 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.\u001b[0m\n",
      "  @numba.jit()\n",
      "c:\\dev\\topic_modelling\\lib\\site-packages\\umap\\distances.py:1086: NumbaDeprecationWarning: \u001b[1mThe 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.\u001b[0m\n",
      "  @numba.jit()\n",
      "c:\\dev\\topic_modelling\\lib\\site-packages\\tqdm\\auto.py:21: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n",
      "  from .autonotebook import tqdm as notebook_tqdm\n",
      "c:\\dev\\topic_modelling\\lib\\site-packages\\umap\\umap_.py:660: NumbaDeprecationWarning: \u001b[1mThe 'nopython' keyword argument was not supplied to the 'numba.jit' decorator. The implicit default value for this argument is currently False, but it will be changed to True in Numba 0.59.0. See https://numba.readthedocs.io/en/stable/reference/deprecation.html#deprecation-of-object-mode-fall-back-behaviour-when-using-jit for details.\u001b[0m\n",
      "  @numba.jit()\n"
     ]
    }
   ],
   "source": [
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "from  tm_common import *\n",
    "IN_COLAB, output_dir = tm_get_working_config()\n",
    "import tensorflow_hub as hub\n",
    "from sklearn.metrics import classification_report\n",
    "from bertopic import BERTopic\n",
    "from bertopic.vectorizers import ClassTfidfTransformer\n",
    "from bertopic.dimensionality import BaseDimensionalityReduction\n",
    "from sklearn.linear_model import LogisticRegression\n",
    "import tensorflow_text\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import (\n",
    "    Flatten,\n",
    "    Embedding,\n",
    "    Dense,\n",
    "    Input,\n",
    "    Embedding,\n",
    "    GlobalAveragePooling1D,\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# load train test sets\n",
    "X_train, X_test, y_train, y_test    = tm_load_train_test_set(output_dir)\n",
    "X_corpus_train, X_corpus_test       = tm_load_train_test_set(output_dir, option=\"raw_corpus\")\n",
    "# # only take a subset to fasten computation \n",
    "for i in range(2):\n",
    "    X_corpus_train, y_train       = tm_get_subset(X_corpus_train, y_train)\n",
    "    X_corpus_test, y_test       = tm_get_subset(X_corpus_test, y_test)\n",
    "label_list = tm_get_label_list(output_dir)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_train_b, y_test_b = tm_multilabel_binarizer(y_train, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1545\n",
      "386\n"
     ]
    }
   ],
   "source": [
    "print( len(X_corpus_train) )\n",
    "print( len(y_test_b)       )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load BERT and the preprocessing model from TF Hub.\n",
    "preprocess = hub.load('https://tfhub.dev/tensorflow/bert_en_uncased_preprocess/1')\n",
    "encoder = hub.load('https://tfhub.dev/tensorflow/bert_en_uncased_L-12_H-768_A-12/3')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_bert_preproc = preprocess(X_corpus_train)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['input_mask', 'input_type_ids', 'input_word_ids'])"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train_bert_preproc.keys()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_bert_emb = encoder(X_train_bert_preproc)['pooled_output']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "del X_train_bert_preproc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_test_bert_preproc = preprocess(X_corpus_test)\n",
    "X_test_bert_emb = encoder(X_test_bert_preproc)['pooled_output']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "del X_test_bert_preproc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " dense (Dense)               (None, 256)               196864    \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 128)               32896     \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 30)                3870      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 233,630\n",
      "Trainable params: 233,630\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "bert_model = Sequential()\n",
    "bert_model.add(Dense(256, activation='relu', input_dim=768))\n",
    "bert_model.add(Dense(128, activation='relu'))\n",
    "bert_model.add(Dense(30,activation='sigmoid'))\n",
    "bert_model.compile(optimizer='adam', loss='binary_crossentropy',metrics=['accuracy'])\n",
    "bert_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "13/13 [==============================] - 1s 3ms/step - loss: 0.3862 - accuracy: 0.0718\n",
      "Epoch 2/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.2222 - accuracy: 0.1353\n",
      "Epoch 3/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1980 - accuracy: 0.1476\n",
      "Epoch 4/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1956 - accuracy: 0.1547\n",
      "Epoch 5/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1934 - accuracy: 0.1592\n",
      "Epoch 6/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1927 - accuracy: 0.1786\n",
      "Epoch 7/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1921 - accuracy: 0.1909\n",
      "Epoch 8/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1917 - accuracy: 0.1612\n",
      "Epoch 9/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1909 - accuracy: 0.1851\n",
      "Epoch 10/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1900 - accuracy: 0.1903\n",
      "Epoch 11/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1899 - accuracy: 0.1981\n",
      "Epoch 12/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1891 - accuracy: 0.2149\n",
      "Epoch 13/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1888 - accuracy: 0.2214\n",
      "Epoch 14/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1870 - accuracy: 0.2155\n",
      "Epoch 15/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1861 - accuracy: 0.2123\n",
      "Epoch 16/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1854 - accuracy: 0.2324\n",
      "Epoch 17/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1840 - accuracy: 0.2395\n",
      "Epoch 18/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1853 - accuracy: 0.2382\n",
      "Epoch 19/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1830 - accuracy: 0.2129\n",
      "Epoch 20/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1830 - accuracy: 0.2220\n",
      "Epoch 21/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1806 - accuracy: 0.2557\n",
      "Epoch 22/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1785 - accuracy: 0.2693\n",
      "Epoch 23/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1772 - accuracy: 0.2757\n",
      "Epoch 24/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1769 - accuracy: 0.2751\n",
      "Epoch 25/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1780 - accuracy: 0.2893\n",
      "Epoch 26/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1760 - accuracy: 0.2913\n",
      "Epoch 27/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1754 - accuracy: 0.2744\n",
      "Epoch 28/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1746 - accuracy: 0.2783\n",
      "Epoch 29/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1735 - accuracy: 0.2945\n",
      "Epoch 30/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1722 - accuracy: 0.2848\n",
      "Epoch 31/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1725 - accuracy: 0.2880\n",
      "Epoch 32/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1723 - accuracy: 0.2913\n",
      "Epoch 33/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1726 - accuracy: 0.2900\n",
      "Epoch 34/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1705 - accuracy: 0.2964\n",
      "Epoch 35/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1697 - accuracy: 0.3049\n",
      "Epoch 36/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1698 - accuracy: 0.3055\n",
      "Epoch 37/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1697 - accuracy: 0.3036\n",
      "Epoch 38/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1679 - accuracy: 0.3010\n",
      "Epoch 39/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1676 - accuracy: 0.3081\n",
      "Epoch 40/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1664 - accuracy: 0.3100\n",
      "Epoch 41/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1685 - accuracy: 0.3055\n",
      "Epoch 42/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1666 - accuracy: 0.3139\n",
      "Epoch 43/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1650 - accuracy: 0.3146\n",
      "Epoch 44/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1662 - accuracy: 0.3133\n",
      "Epoch 45/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1645 - accuracy: 0.3230\n",
      "Epoch 46/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1677 - accuracy: 0.3327\n",
      "Epoch 47/100\n",
      "13/13 [==============================] - 0s 4ms/step - loss: 0.1654 - accuracy: 0.3178\n",
      "Epoch 48/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1680 - accuracy: 0.3029\n",
      "Epoch 49/100\n",
      "13/13 [==============================] - 0s 4ms/step - loss: 0.1651 - accuracy: 0.3087\n",
      "Epoch 50/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1634 - accuracy: 0.3346\n",
      "Epoch 51/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1640 - accuracy: 0.3314\n",
      "Epoch 52/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1611 - accuracy: 0.3430\n",
      "Epoch 53/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1605 - accuracy: 0.3437\n",
      "Epoch 54/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1593 - accuracy: 0.3515\n",
      "Epoch 55/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1595 - accuracy: 0.3456\n",
      "Epoch 56/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1601 - accuracy: 0.3320\n",
      "Epoch 57/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1602 - accuracy: 0.3430\n",
      "Epoch 58/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1582 - accuracy: 0.3476\n",
      "Epoch 59/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1599 - accuracy: 0.3191\n",
      "Epoch 60/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1575 - accuracy: 0.3663\n",
      "Epoch 61/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1554 - accuracy: 0.3592\n",
      "Epoch 62/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1581 - accuracy: 0.3294\n",
      "Epoch 63/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1571 - accuracy: 0.3553\n",
      "Epoch 64/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1598 - accuracy: 0.3359\n",
      "Epoch 65/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1545 - accuracy: 0.3864\n",
      "Epoch 66/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1545 - accuracy: 0.3676\n",
      "Epoch 67/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1540 - accuracy: 0.3883\n",
      "Epoch 68/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1537 - accuracy: 0.3735\n",
      "Epoch 69/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1547 - accuracy: 0.3754\n",
      "Epoch 70/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1531 - accuracy: 0.3696\n",
      "Epoch 71/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1518 - accuracy: 0.3612\n",
      "Epoch 72/100\n",
      "13/13 [==============================] - 0s 2ms/step - loss: 0.1497 - accuracy: 0.4032\n",
      "Epoch 73/100\n",
      "13/13 [==============================] - 0s 2ms/step - loss: 0.1496 - accuracy: 0.3864\n",
      "Epoch 74/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1496 - accuracy: 0.3858\n",
      "Epoch 75/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1488 - accuracy: 0.4065\n",
      "Epoch 76/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1486 - accuracy: 0.4175\n",
      "Epoch 77/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1484 - accuracy: 0.3916\n",
      "Epoch 78/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1475 - accuracy: 0.4129\n",
      "Epoch 79/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1482 - accuracy: 0.4110\n",
      "Epoch 80/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1475 - accuracy: 0.4071\n",
      "Epoch 81/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1473 - accuracy: 0.4026\n",
      "Epoch 82/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1462 - accuracy: 0.3961\n",
      "Epoch 83/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1466 - accuracy: 0.3968\n",
      "Epoch 84/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1452 - accuracy: 0.4149\n",
      "Epoch 85/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1449 - accuracy: 0.4058\n",
      "Epoch 86/100\n",
      "13/13 [==============================] - 0s 2ms/step - loss: 0.1441 - accuracy: 0.4291\n",
      "Epoch 87/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1421 - accuracy: 0.4324\n",
      "Epoch 88/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1439 - accuracy: 0.4142\n",
      "Epoch 89/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1435 - accuracy: 0.4155\n",
      "Epoch 90/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1460 - accuracy: 0.3909\n",
      "Epoch 91/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1417 - accuracy: 0.4285\n",
      "Epoch 92/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1409 - accuracy: 0.4317\n",
      "Epoch 93/100\n",
      "13/13 [==============================] - 0s 2ms/step - loss: 0.1443 - accuracy: 0.4084\n",
      "Epoch 94/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1478 - accuracy: 0.3994\n",
      "Epoch 95/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1416 - accuracy: 0.4278\n",
      "Epoch 96/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1423 - accuracy: 0.3909\n",
      "Epoch 97/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1396 - accuracy: 0.4350\n",
      "Epoch 98/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1390 - accuracy: 0.4343\n",
      "Epoch 99/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1396 - accuracy: 0.4239\n",
      "Epoch 100/100\n",
      "13/13 [==============================] - 0s 3ms/step - loss: 0.1397 - accuracy: 0.4123\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x2265c2eb730>"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "bert_model.fit(X_train_bert_emb, y_train_b, batch_size = 128, epochs=100)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4/4 [==============================] - 0s 2ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred = bert_model.predict(X_train_bert_emb[:100])\n",
    "threshold_test_use = tm_test_threshold(y_train_b[:100], y_pred)  \n",
    "tm_plot_threshold_test(threshold_test_use)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.25118864315094075"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max_config = threshold_test_use.query('precision == precision.max()')\n",
    "t = max_config[\"threshold\"]\n",
    "best_t =  t.values[0]\n",
    "best_t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "13/13 [==============================] - 0s 2ms/step\n"
     ]
    }
   ],
   "source": [
    "y_pred_bert = bert_model.predict(X_test_bert_emb)\n",
    "y_pred_bert = (y_pred_bert > best_t).astype(np.float32)\n",
    "precision_bert      = average_precision_score(y_test_b, y_pred_bert, average='micro')\n",
    "jaccard_score_bert = jaccard_score(y_test_b, y_pred_bert, average='micro')\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 3000x1000 with 3 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(y_test_b, y_pred_bert, label_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "                precision    recall  f1-score   support\n",
      "\n",
      "            c#       0.29      0.43      0.35        58\n",
      "          java       0.48      0.37      0.42        68\n",
      "           c++       0.37      0.65      0.47        48\n",
      "    javascript       0.35      0.64      0.45        45\n",
      "        python       0.66      0.46      0.54        50\n",
      "       android       0.50      0.48      0.49        31\n",
      "           ios       0.53      0.26      0.35        35\n",
      "          .net       0.00      0.00      0.00        29\n",
      "           php       0.60      0.33      0.43        18\n",
      "             c       0.25      0.18      0.21        22\n",
      "          html       0.26      0.50      0.34        18\n",
      "        jquery       0.29      0.31      0.30        16\n",
      "   performance       0.00      0.00      0.00        16\n",
      "           css       0.40      0.12      0.19        16\n",
      "   objective-c       0.00      0.00      0.00        12\n",
      "         c++11       0.29      0.17      0.21        12\n",
      "       asp.net       0.00      0.00      0.00         7\n",
      "       node.js       0.00      0.00      0.00        11\n",
      "           sql       0.00      0.00      0.00         9\n",
      "     algorithm       0.50      0.10      0.17        10\n",
      "         linux       0.00      0.00      0.00         8\n",
      "        spring       0.00      0.00      0.00         9\n",
      "multithreading       0.17      0.14      0.15         7\n",
      "        iphone       0.00      0.00      0.00         9\n",
      "         swift       0.50      0.08      0.14        12\n",
      "         mysql       0.00      0.00      0.00        12\n",
      "       windows       0.00      0.00      0.00         7\n",
      "          json       0.00      0.00      0.00         5\n",
      "        arrays       0.00      0.00      0.00         8\n",
      "         xcode       0.00      0.00      0.00         8\n",
      "\n",
      "     micro avg       0.37      0.31      0.34       616\n",
      "     macro avg       0.21      0.17      0.17       616\n",
      "  weighted avg       0.32      0.31      0.29       616\n",
      "   samples avg       0.32      0.33      0.30       616\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(y_test_b, y_pred_bert, target_names=label_list, digits=2))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here, the support is clearly below the one I used for other algorithm.  It explains the lack of precision for very under-represented classes in the test set. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "avg_precision_scores = [precision_bert]\n",
    "jaccard_scores = [jaccard_score_bert]\n",
    "result = pd.DataFrame(\n",
    "    {\n",
    "        'average_precision_scores' : avg_precision_scores,\n",
    "        'jaccard_scores' : jaccard_scores\n",
    "    }, index = ['BERT']\n",
    ")\n",
    "result.to_csv(f\"{output_dir}result_supervised3.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "result1 = pd.read_csv(f\"{output_dir}/supervised_1_results.csv\")\n",
    "result2 = pd.read_csv(f\"{output_dir}/result_supervised2.csv\")\n",
    "result3 = pd.read_csv(f\"{output_dir}/result_supervised3.csv\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Results"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "USE embedding is the one achieving the best precision. The SVM algorithm also achieve close performances.  This algorithm  is computationally  less intensive. This could be a good trade off worth taking into consideration. For further work, it would be interesting to train these models on larger dataset especially BERT one. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model</th>\n",
       "      <th>average_precision_scores</th>\n",
       "      <th>jaccard_scores</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Dummy classifier</td>\n",
       "      <td>0.053830</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>logistic : concatenation</td>\n",
       "      <td>0.196039</td>\n",
       "      <td>0.170827</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>SVM : concatenation</td>\n",
       "      <td>0.422422</td>\n",
       "      <td>0.434641</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>RandomForest</td>\n",
       "      <td>0.321323</td>\n",
       "      <td>0.372925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Word2Vec</td>\n",
       "      <td>0.085256</td>\n",
       "      <td>0.112318</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>USE</td>\n",
       "      <td>0.463946</td>\n",
       "      <td>0.500000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>BERT</td>\n",
       "      <td>0.150803</td>\n",
       "      <td>0.201717</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                      model  average_precision_scores  jaccard_scores\n",
       "0          Dummy classifier                  0.053830        0.000000\n",
       "1  logistic : concatenation                  0.196039        0.170827\n",
       "2      SVM : concatenation                   0.422422        0.434641\n",
       "3              RandomForest                  0.321323        0.372925\n",
       "0                  Word2Vec                  0.085256        0.112318\n",
       "1                       USE                  0.463946        0.500000\n",
       "0                      BERT                  0.150803        0.201717"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "supervised_restult = pd.concat([result1, result2,result3], axis=0)\n",
    "supervised_restult.rename(columns={'Unnamed: 0' : 'model'})"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "topic_modelling",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.10"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}