word2vec_with_LMDB.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 597,
   "id": "fa5d3348",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     C:\\Users\\kojun\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import re\n",
    "from tqdm import tqdm\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import nltk\n",
    "nltk.download(\"stopwords\")\n",
    "from nltk.corpus import stopwords\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.utils import shuffle\n",
    "from sklearn.metrics import accuracy_score,confusion_matrix\n",
    "\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.layers import Flatten,Dense,Dropout,BatchNormalization\n",
    "from tensorflow.keras.regularizers import l2\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from gensim.models.word2vec import Word2Vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 550,
   "id": "78c0c457",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_neg_list = os.listdir('LMDB/aclImdb_v1/aclImdb/train/neg')\n",
    "train_pos_list = os.listdir('LMDB/aclImdb_v1/aclImdb/train/pos')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 551,
   "id": "6d6bf69b",
   "metadata": {},
   "outputs": [],
   "source": [
    "pos_path = 'C:/Users/kojun/Desktop/ML/deep learning/文字/LMDB/aclImdb_v1/aclImdb/train/pos/'\n",
    "neg_path = 'C:/Users/kojun/Desktop/ML/deep learning/文字/LMDB/aclImdb_v1/aclImdb/train/neg/'"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "72e4d875",
   "metadata": {},
   "source": [
    "# POS"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 552,
   "id": "a8befc2b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████| 12500/12500 [00:00<00:00, 33418.70it/s]\n"
     ]
    }
   ],
   "source": [
    "one=[]\n",
    "pos_score = []\n",
    "pos_content = []\n",
    "for i in tqdm(range(len(train_pos_list))):\n",
    "    with open(os.path.join(pos_path,train_pos_list[i]),'r',encoding=\"utf-8\") as f:\n",
    "        pos_score.append(int(train_pos_list[i].split('.')[0].split('_')[1]))\n",
    "        pos_content.append(f.read())\n",
    "        one.append('pos')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "418749fd",
   "metadata": {},
   "source": [
    "# Neg"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 553,
   "id": "b1b25843",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████| 12500/12500 [00:00<00:00, 31276.46it/s]\n"
     ]
    }
   ],
   "source": [
    "zero=[]\n",
    "neg_score = []\n",
    "neg_content = []\n",
    "for i in tqdm(range(len(train_pos_list))):\n",
    "    with open(os.path.join(neg_path,train_neg_list[i]),'r',encoding=\"utf-8\") as f:\n",
    "        neg_score.append(int(train_neg_list[i].split('.')[0].split('_')[1]))\n",
    "        neg_content.append(f.read())\n",
    "        zero.append('neg')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "6afd1b00",
   "metadata": {},
   "source": [
    "# Concat"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 554,
   "id": "4f0c547d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4"
      ]
     },
     "execution_count": 554,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "max(neg_score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 555,
   "id": "1cd65d03",
   "metadata": {},
   "outputs": [],
   "source": [
    "call_content = pos_content+neg_content\n",
    "call_score = pos_score+neg_score\n",
    "call_one_zero = one+zero"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 556,
   "id": "99122dd0",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.DataFrame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 557,
   "id": "91f7aee2",
   "metadata": {},
   "outputs": [],
   "source": [
    "train['Contain'] = call_content\n",
    "train['score'] = call_score\n",
    "train['ps_ne'] = call_one_zero"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 558,
   "id": "95b9b9ed",
   "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>Contain</th>\n",
       "      <th>score</th>\n",
       "      <th>ps_ne</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Bromwell High is a cartoon comedy. It ran at t...</td>\n",
       "      <td>9</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Homelessness (or Houselessness as George Carli...</td>\n",
       "      <td>8</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Brilliant over-acting by Lesley Ann Warren. Be...</td>\n",
       "      <td>10</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>This is easily the most underrated film inn th...</td>\n",
       "      <td>7</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>This is not the typical Mel Brooks film. It wa...</td>\n",
       "      <td>8</td>\n",
       "      <td>pos</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24995</th>\n",
       "      <td>Towards the end of the movie, I felt it was to...</td>\n",
       "      <td>4</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24996</th>\n",
       "      <td>This is the kind of movie that my enemies cont...</td>\n",
       "      <td>3</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24997</th>\n",
       "      <td>I saw 'Descent' last night at the Stockholm Fi...</td>\n",
       "      <td>3</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24998</th>\n",
       "      <td>Some films that you pick up for a pound turn o...</td>\n",
       "      <td>1</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24999</th>\n",
       "      <td>This is one of the dumbest films, I've ever se...</td>\n",
       "      <td>1</td>\n",
       "      <td>neg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25000 rows × 3 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 Contain  score ps_ne\n",
       "0      Bromwell High is a cartoon comedy. It ran at t...      9   pos\n",
       "1      Homelessness (or Houselessness as George Carli...      8   pos\n",
       "2      Brilliant over-acting by Lesley Ann Warren. Be...     10   pos\n",
       "3      This is easily the most underrated film inn th...      7   pos\n",
       "4      This is not the typical Mel Brooks film. It wa...      8   pos\n",
       "...                                                  ...    ...   ...\n",
       "24995  Towards the end of the movie, I felt it was to...      4   neg\n",
       "24996  This is the kind of movie that my enemies cont...      3   neg\n",
       "24997  I saw 'Descent' last night at the Stockholm Fi...      3   neg\n",
       "24998  Some films that you pick up for a pound turn o...      1   neg\n",
       "24999  This is one of the dumbest films, I've ever se...      1   neg\n",
       "\n",
       "[25000 rows x 3 columns]"
      ]
     },
     "execution_count": 558,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "993052a7",
   "metadata": {},
   "source": [
    "# data clean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 559,
   "id": "e57b4f50",
   "metadata": {},
   "outputs": [],
   "source": [
    "def text_cleaning(text):\n",
    "    forbidden_words = set(stopwords.words('english'))\n",
    "    if text:\n",
    "        text = ' '.join(text.split('.'))\n",
    "        text = re.sub('\\/',' ',text)\n",
    "        text = re.sub(r'\\\\',' ',text)\n",
    "        text = re.sub(r'((http)\\S+)','',text)\n",
    "        text = re.sub(r'\\s+', ' ', re.sub('[^A-Za-z]', ' ', text.strip().lower())).strip()\n",
    "        text = re.sub(r'\\W+', ' ', text.strip().lower()).strip()\n",
    "        text = [word for word in text.split() if word not in forbidden_words]\n",
    "        return \" \".join(text)\n",
    "    return \"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 560,
   "id": "6d067f33",
   "metadata": {},
   "outputs": [],
   "source": [
    "train['Contain'] = train.apply(lambda row:text_cleaning(row['Contain']),axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 561,
   "id": "796d49c7",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_all = train[['Contain','ps_ne']]"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "0f10ea92",
   "metadata": {},
   "source": [
    "# word2vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 562,
   "id": "ebb9bb9b",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\kojun\\Anaconda3\\lib\\site-packages\\pandas\\core\\indexing.py:1732: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  self._setitem_single_block(indexer, value, name)\n",
      "C:\\Users\\kojun\\Anaconda3\\lib\\site-packages\\pandas\\core\\indexing.py:723: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: https://pandas.pydata.org/pandas-docs/stable/user_guide/indexing.html#returning-a-view-versus-a-copy\n",
      "  iloc._setitem_with_indexer(indexer, value, self.name)\n"
     ]
    }
   ],
   "source": [
    "input_all.loc[input_all['ps_ne']=='pos','ps_ne'] = 1\n",
    "input_all.loc[input_all['ps_ne']=='neg','ps_ne'] = 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 563,
   "id": "349c451d",
   "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>Contain</th>\n",
       "      <th>ps_ne</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>bromwell high cartoon comedy ran time programs...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>homelessness houselessness george carlin state...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>brilliant acting lesley ann warren best dramat...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>easily underrated film inn brooks cannon sure ...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>typical mel brooks film much less slapstick mo...</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24995</th>\n",
       "      <td>towards end movie felt technical felt like cla...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24996</th>\n",
       "      <td>kind movie enemies content watch time bloody t...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24997</th>\n",
       "      <td>saw descent last night stockholm film festival...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24998</th>\n",
       "      <td>films pick pound turn rather good rd century f...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24999</th>\n",
       "      <td>one dumbest films ever seen rips nearly ever t...</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>25000 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                 Contain ps_ne\n",
       "0      bromwell high cartoon comedy ran time programs...     1\n",
       "1      homelessness houselessness george carlin state...     1\n",
       "2      brilliant acting lesley ann warren best dramat...     1\n",
       "3      easily underrated film inn brooks cannon sure ...     1\n",
       "4      typical mel brooks film much less slapstick mo...     1\n",
       "...                                                  ...   ...\n",
       "24995  towards end movie felt technical felt like cla...     0\n",
       "24996  kind movie enemies content watch time bloody t...     0\n",
       "24997  saw descent last night stockholm film festival...     0\n",
       "24998  films pick pound turn rather good rd century f...     0\n",
       "24999  one dumbest films ever seen rips nearly ever t...     0\n",
       "\n",
       "[25000 rows x 2 columns]"
      ]
     },
     "execution_count": 563,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "input_all"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 564,
   "id": "a3a1658f",
   "metadata": {},
   "outputs": [],
   "source": [
    "vector_size = 400"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 565,
   "id": "4d9d9c2b",
   "metadata": {},
   "outputs": [],
   "source": [
    "w2v_model = Word2Vec(input_all['Contain'].apply(lambda x: x.split(' ')),vector_size=vector_size,sg=1,min_count=100,workers=20,epochs=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 566,
   "id": "93b65bf4",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'bromwell high cartoon comedy ran time programs school life teachers years teaching profession lead believe bromwell high satire much closer reality teachers scramble survive financially insightful students see right pathetic teachers pomp pettiness whole situation remind schools knew students saw episode student repeatedly tried burn school immediately recalled high classic line inspector sack one teachers student welcome bromwell high expect many adults age think bromwell high far fetched pity'"
      ]
     },
     "execution_count": 566,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "input_all.loc[0,'Contain']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 567,
   "id": "823b7cf1",
   "metadata": {},
   "outputs": [],
   "source": [
    "# vec=[]\n",
    "# for word in input_all.loc[0,'Contain'].split(' '):\n",
    "#     vec.append(w2v_model.wv[word])\n",
    "# vec = np.array(vec)\n",
    "# vec = np.mean(vec,axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 568,
   "id": "9aa734cc",
   "metadata": {},
   "outputs": [],
   "source": [
    "def average_verc(df):\n",
    "    vec =[]\n",
    "    for word in df.split(' '):\n",
    "        if word in w2v_model.wv.index_to_key:\n",
    "            vec.append(w2v_model.wv[word])\n",
    "        else:\n",
    "            pass\n",
    "    vec = np.array(vec)\n",
    "    vec = np.mean(vec,axis=0)\n",
    "    return vec"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 569,
   "id": "5f878b08",
   "metadata": {},
   "outputs": [],
   "source": [
    "test = input_all.apply(lambda row:average_verc(row['Contain']),axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 570,
   "id": "c5240cc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "g= np.empty(shape=(25000,vector_size))\n",
    "for i in range(len(test)):\n",
    "    g[i] = test[i]\n",
    "g=pd.DataFrame(g)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 571,
   "id": "0b5d1fc1",
   "metadata": {},
   "outputs": [],
   "source": [
    "g['label'] = input_all['ps_ne']"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "f85bae75",
   "metadata": {},
   "source": [
    "# train / vali split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 572,
   "id": "3f609c68",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(50):\n",
    "    g=shuffle(g)\n",
    "g = g.reset_index(drop=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 573,
   "id": "cbec23f6",
   "metadata": {},
   "outputs": [],
   "source": [
    "input_x = g.drop('label',axis=1)\n",
    "input_y = g['label']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 574,
   "id": "b3f5fc40",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_x,valid_x,train_y,valid_y = train_test_split(input_x,input_y,test_size=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 575,
   "id": "5cc41566",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((20000, 400), (5000, 400), (20000,), (5000,))"
      ]
     },
     "execution_count": 575,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_x.shape,valid_x.shape,train_y.shape,valid_y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 576,
   "id": "cecc7f0c",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_y = train_y.astype('float16')\n",
    "valid_y = valid_y.astype('float16')"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "524a70e2",
   "metadata": {},
   "source": [
    "# Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 658,
   "id": "3b5c696d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_22\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " flatten_22 (Flatten)        (None, 400)               0         \n",
      "                                                                 \n",
      " dense_61 (Dense)            (None, 50)                20050     \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 50)               200       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " dense_62 (Dense)            (None, 2)                 102       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 20,352\n",
      "Trainable params: 20,252\n",
      "Non-trainable params: 100\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "req = 1e-2\n",
    "#,kernel_regularizer=l2(req),bias_regularizer=l2(req)\n",
    "DNN_model = tf.keras.models.Sequential([\n",
    "    #input layer\n",
    "    Flatten(input_shape=(vector_size,)),\n",
    "    #hidden layer\n",
    "    Dense(units=50,activation='relu',kernel_initializer='he_uniform',kernel_regularizer=l2(req),bias_regularizer=l2(req)),\n",
    "    BatchNormalization(),\n",
    "    #output layer\n",
    "    Dense(units=2,activation='sigmoid') #softmax\n",
    "])\n",
    "\n",
    "DNN_model.summary()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 678,
   "id": "5b51cc7a",
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = tf.keras.optimizers.Adam(learning_rate=1e-3)\n",
    "\n",
    "loss = tf.keras.losses.SparseCategoricalCrossentropy()\n",
    "\n",
    "DNN_model.compile(optimizer=optimizer ,loss=loss ,metrics=['accuracy'])\n",
    "\n",
    "epock=80\n",
    "batch_size= 80\n",
    "\n",
    "\n",
    "def scheler(epoch,lr):\n",
    "    if epoch < 25:\n",
    "        return lr\n",
    "    else:\n",
    "        return lr*np.exp(-0.1)\n",
    "    \n",
    "call = tf.keras.callbacks.LearningRateScheduler(scheler)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 679,
   "id": "feab3ba3",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3324 - accuracy: 0.8621 - val_loss: 0.4324 - val_accuracy: 0.8232 - lr: 0.0010\n",
      "Epoch 2/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3339 - accuracy: 0.8609 - val_loss: 0.3321 - val_accuracy: 0.8648 - lr: 0.0010\n",
      "Epoch 3/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3271 - accuracy: 0.8590 - val_loss: 0.3279 - val_accuracy: 0.8688 - lr: 0.0010\n",
      "Epoch 4/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3241 - accuracy: 0.8651 - val_loss: 0.5136 - val_accuracy: 0.7830 - lr: 0.0010\n",
      "Epoch 5/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3335 - accuracy: 0.8619 - val_loss: 0.3358 - val_accuracy: 0.8708 - lr: 0.0010\n",
      "Epoch 6/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3253 - accuracy: 0.8637 - val_loss: 0.3377 - val_accuracy: 0.8640 - lr: 0.0010\n",
      "Epoch 7/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3238 - accuracy: 0.8655 - val_loss: 0.3347 - val_accuracy: 0.8696 - lr: 0.0010\n",
      "Epoch 8/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3243 - accuracy: 0.8668 - val_loss: 0.3507 - val_accuracy: 0.8616 - lr: 0.0010\n",
      "Epoch 9/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3269 - accuracy: 0.8663 - val_loss: 0.5716 - val_accuracy: 0.7662 - lr: 0.0010\n",
      "Epoch 10/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3276 - accuracy: 0.8655 - val_loss: 0.3843 - val_accuracy: 0.8394 - lr: 0.0010\n",
      "Epoch 11/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3270 - accuracy: 0.8653 - val_loss: 0.3569 - val_accuracy: 0.8540 - lr: 0.0010\n",
      "Epoch 12/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3247 - accuracy: 0.8661 - val_loss: 0.3361 - val_accuracy: 0.8708 - lr: 0.0010\n",
      "Epoch 13/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3234 - accuracy: 0.8671 - val_loss: 0.3397 - val_accuracy: 0.8696 - lr: 0.0010\n",
      "Epoch 14/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3269 - accuracy: 0.8653 - val_loss: 0.4073 - val_accuracy: 0.8294 - lr: 0.0010\n",
      "Epoch 15/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3288 - accuracy: 0.8665 - val_loss: 0.3379 - val_accuracy: 0.8650 - lr: 0.0010\n",
      "Epoch 16/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3325 - accuracy: 0.8629 - val_loss: 0.3759 - val_accuracy: 0.8436 - lr: 0.0010\n",
      "Epoch 17/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3268 - accuracy: 0.8652 - val_loss: 0.4608 - val_accuracy: 0.8068 - lr: 0.0010\n",
      "Epoch 18/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3242 - accuracy: 0.8668 - val_loss: 0.3462 - val_accuracy: 0.8654 - lr: 0.0010\n",
      "Epoch 19/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3231 - accuracy: 0.8679 - val_loss: 0.3323 - val_accuracy: 0.8724 - lr: 0.0010\n",
      "Epoch 20/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3271 - accuracy: 0.8670 - val_loss: 0.3308 - val_accuracy: 0.8726 - lr: 0.0010\n",
      "Epoch 21/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3219 - accuracy: 0.8686 - val_loss: 0.3440 - val_accuracy: 0.8678 - lr: 0.0010\n",
      "Epoch 22/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3230 - accuracy: 0.8687 - val_loss: 0.3525 - val_accuracy: 0.8616 - lr: 0.0010\n",
      "Epoch 23/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3216 - accuracy: 0.8685 - val_loss: 0.3340 - val_accuracy: 0.8720 - lr: 0.0010\n",
      "Epoch 24/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3221 - accuracy: 0.8675 - val_loss: 0.3858 - val_accuracy: 0.8460 - lr: 0.0010\n",
      "Epoch 25/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3227 - accuracy: 0.8676 - val_loss: 0.3932 - val_accuracy: 0.8402 - lr: 0.0010\n",
      "Epoch 26/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3270 - accuracy: 0.8665 - val_loss: 0.3297 - val_accuracy: 0.8742 - lr: 9.0484e-04\n",
      "Epoch 27/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3201 - accuracy: 0.8690 - val_loss: 0.3525 - val_accuracy: 0.8608 - lr: 8.1873e-04\n",
      "Epoch 28/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3191 - accuracy: 0.8704 - val_loss: 0.3479 - val_accuracy: 0.8614 - lr: 7.4082e-04\n",
      "Epoch 29/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3170 - accuracy: 0.8698 - val_loss: 0.3308 - val_accuracy: 0.8726 - lr: 6.7032e-04\n",
      "Epoch 30/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3121 - accuracy: 0.8731 - val_loss: 0.3977 - val_accuracy: 0.8356 - lr: 6.0653e-04\n",
      "Epoch 31/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3132 - accuracy: 0.8712 - val_loss: 0.3421 - val_accuracy: 0.8644 - lr: 5.4881e-04\n",
      "Epoch 32/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3105 - accuracy: 0.8725 - val_loss: 0.3406 - val_accuracy: 0.8668 - lr: 4.9659e-04\n",
      "Epoch 33/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3110 - accuracy: 0.8737 - val_loss: 0.3335 - val_accuracy: 0.8680 - lr: 4.4933e-04\n",
      "Epoch 34/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3083 - accuracy: 0.8741 - val_loss: 0.3318 - val_accuracy: 0.8686 - lr: 4.0657e-04\n",
      "Epoch 35/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3050 - accuracy: 0.8760 - val_loss: 0.3284 - val_accuracy: 0.8706 - lr: 3.6788e-04\n",
      "Epoch 36/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3045 - accuracy: 0.8755 - val_loss: 0.3360 - val_accuracy: 0.8670 - lr: 3.3287e-04\n",
      "Epoch 37/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3030 - accuracy: 0.8763 - val_loss: 0.3336 - val_accuracy: 0.8702 - lr: 3.0119e-04\n",
      "Epoch 38/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3014 - accuracy: 0.8776 - val_loss: 0.3366 - val_accuracy: 0.8688 - lr: 2.7253e-04\n",
      "Epoch 39/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.3001 - accuracy: 0.8799 - val_loss: 0.3279 - val_accuracy: 0.8710 - lr: 2.4660e-04\n",
      "Epoch 40/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.3000 - accuracy: 0.8778 - val_loss: 0.3304 - val_accuracy: 0.8702 - lr: 2.2313e-04\n",
      "Epoch 41/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2979 - accuracy: 0.8787 - val_loss: 0.3312 - val_accuracy: 0.8696 - lr: 2.0190e-04\n",
      "Epoch 42/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2985 - accuracy: 0.8785 - val_loss: 0.3323 - val_accuracy: 0.8694 - lr: 1.8268e-04\n",
      "Epoch 43/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2977 - accuracy: 0.8803 - val_loss: 0.3289 - val_accuracy: 0.8738 - lr: 1.6530e-04\n",
      "Epoch 44/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2966 - accuracy: 0.8781 - val_loss: 0.3280 - val_accuracy: 0.8728 - lr: 1.4957e-04\n",
      "Epoch 45/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2938 - accuracy: 0.8812 - val_loss: 0.3289 - val_accuracy: 0.8708 - lr: 1.3534e-04\n",
      "Epoch 46/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2937 - accuracy: 0.8808 - val_loss: 0.3310 - val_accuracy: 0.8702 - lr: 1.2246e-04\n",
      "Epoch 47/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2951 - accuracy: 0.8798 - val_loss: 0.3335 - val_accuracy: 0.8688 - lr: 1.1080e-04\n",
      "Epoch 48/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2939 - accuracy: 0.8808 - val_loss: 0.3284 - val_accuracy: 0.8694 - lr: 1.0026e-04\n",
      "Epoch 49/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2936 - accuracy: 0.8809 - val_loss: 0.3276 - val_accuracy: 0.8698 - lr: 9.0718e-05\n",
      "Epoch 50/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2924 - accuracy: 0.8833 - val_loss: 0.3278 - val_accuracy: 0.8714 - lr: 8.2085e-05\n",
      "Epoch 51/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2930 - accuracy: 0.8810 - val_loss: 0.3278 - val_accuracy: 0.8690 - lr: 7.4274e-05\n",
      "Epoch 52/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2922 - accuracy: 0.8806 - val_loss: 0.3292 - val_accuracy: 0.8704 - lr: 6.7206e-05\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 53/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2910 - accuracy: 0.8828 - val_loss: 0.3299 - val_accuracy: 0.8704 - lr: 6.0810e-05\n",
      "Epoch 54/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2924 - accuracy: 0.8824 - val_loss: 0.3275 - val_accuracy: 0.8688 - lr: 5.5023e-05\n",
      "Epoch 55/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2906 - accuracy: 0.8812 - val_loss: 0.3306 - val_accuracy: 0.8696 - lr: 4.9787e-05\n",
      "Epoch 56/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2927 - accuracy: 0.8808 - val_loss: 0.3279 - val_accuracy: 0.8708 - lr: 4.5049e-05\n",
      "Epoch 57/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2888 - accuracy: 0.8837 - val_loss: 0.3279 - val_accuracy: 0.8694 - lr: 4.0762e-05\n",
      "Epoch 58/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2901 - accuracy: 0.8827 - val_loss: 0.3281 - val_accuracy: 0.8706 - lr: 3.6883e-05\n",
      "Epoch 59/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2906 - accuracy: 0.8816 - val_loss: 0.3280 - val_accuracy: 0.8700 - lr: 3.3373e-05\n",
      "Epoch 60/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2908 - accuracy: 0.8830 - val_loss: 0.3281 - val_accuracy: 0.8690 - lr: 3.0197e-05\n",
      "Epoch 61/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2900 - accuracy: 0.8829 - val_loss: 0.3279 - val_accuracy: 0.8682 - lr: 2.7324e-05\n",
      "Epoch 62/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2896 - accuracy: 0.8810 - val_loss: 0.3278 - val_accuracy: 0.8676 - lr: 2.4724e-05\n",
      "Epoch 63/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2890 - accuracy: 0.8831 - val_loss: 0.3279 - val_accuracy: 0.8684 - lr: 2.2371e-05\n",
      "Epoch 64/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2909 - accuracy: 0.8839 - val_loss: 0.3283 - val_accuracy: 0.8704 - lr: 2.0242e-05\n",
      "Epoch 65/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2876 - accuracy: 0.8817 - val_loss: 0.3282 - val_accuracy: 0.8700 - lr: 1.8316e-05\n",
      "Epoch 66/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2910 - accuracy: 0.8814 - val_loss: 0.3282 - val_accuracy: 0.8706 - lr: 1.6573e-05\n",
      "Epoch 67/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2900 - accuracy: 0.8828 - val_loss: 0.3282 - val_accuracy: 0.8700 - lr: 1.4996e-05\n",
      "Epoch 68/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2898 - accuracy: 0.8827 - val_loss: 0.3282 - val_accuracy: 0.8700 - lr: 1.3569e-05\n",
      "Epoch 69/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2898 - accuracy: 0.8824 - val_loss: 0.3281 - val_accuracy: 0.8700 - lr: 1.2277e-05\n",
      "Epoch 70/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2889 - accuracy: 0.8835 - val_loss: 0.3278 - val_accuracy: 0.8696 - lr: 1.1109e-05\n",
      "Epoch 71/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2886 - accuracy: 0.8820 - val_loss: 0.3280 - val_accuracy: 0.8684 - lr: 1.0052e-05\n",
      "Epoch 72/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2896 - accuracy: 0.8813 - val_loss: 0.3283 - val_accuracy: 0.8706 - lr: 9.0953e-06\n",
      "Epoch 73/80\n",
      "250/250 [==============================] - 1s 5ms/step - loss: 0.2890 - accuracy: 0.8835 - val_loss: 0.3281 - val_accuracy: 0.8696 - lr: 8.2297e-06\n",
      "Epoch 74/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2875 - accuracy: 0.8859 - val_loss: 0.3280 - val_accuracy: 0.8684 - lr: 7.4466e-06\n",
      "Epoch 75/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2889 - accuracy: 0.8814 - val_loss: 0.3281 - val_accuracy: 0.8688 - lr: 6.7379e-06\n",
      "Epoch 76/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2889 - accuracy: 0.8835 - val_loss: 0.3281 - val_accuracy: 0.8696 - lr: 6.0967e-06\n",
      "Epoch 77/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2890 - accuracy: 0.8824 - val_loss: 0.3281 - val_accuracy: 0.8686 - lr: 5.5166e-06\n",
      "Epoch 78/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2880 - accuracy: 0.8842 - val_loss: 0.3280 - val_accuracy: 0.8682 - lr: 4.9916e-06\n",
      "Epoch 79/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2883 - accuracy: 0.8813 - val_loss: 0.3285 - val_accuracy: 0.8698 - lr: 4.5166e-06\n",
      "Epoch 80/80\n",
      "250/250 [==============================] - 1s 4ms/step - loss: 0.2886 - accuracy: 0.8834 - val_loss: 0.3281 - val_accuracy: 0.8684 - lr: 4.0868e-06\n"
     ]
    }
   ],
   "source": [
    "History = DNN_model.fit(x=train_x, \n",
    "                        y=train_y,\n",
    "                        batch_size=batch_size,\n",
    "                        epochs=epock,\n",
    "                        validation_data=(valid_x,valid_y),\n",
    "                        callbacks=[call])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 680,
   "id": "ed35415f",
   "metadata": {},
   "outputs": [],
   "source": [
    "loss = History.history[\"loss\"]\n",
    "accuracy = History.history[\"accuracy\"]\n",
    "val_loss = History.history[\"val_loss\"]\n",
    "val_accuracy = History.history[\"val_accuracy\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 681,
   "id": "136d4018",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# loss\n",
    "plt.figure(figsize=(10,6))\n",
    "plt.plot(loss, label=\"train loss\")\n",
    "plt.plot(val_loss, label=\"val loss\")\n",
    "plt.title(\"Loss\", fontsize=20)\n",
    "plt.xlabel(\"epoch\", fontsize=16)\n",
    "plt.ylabel(\"loss\", fontsize=16)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 682,
   "id": "3cc61a05",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# accuracy\n",
    "plt.figure(figsize=(10,6))\n",
    "plt.plot(accuracy, label=\"train acc\")\n",
    "plt.plot(val_accuracy, label=\"val acc\")\n",
    "plt.title(\"Accuracy\", fontsize=20)\n",
    "plt.xlabel(\"epoch\", fontsize=16)\n",
    "plt.ylabel(\"accuracy\", fontsize=16)\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "id": "17a3bfc2",
   "metadata": {},
   "source": [
    "# test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 683,
   "id": "ee6ff9ef",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_neg_list = os.listdir('LMDB/aclImdb_v1/aclImdb/test/neg')\n",
    "test_pos_list = os.listdir('LMDB/aclImdb_v1/aclImdb/test/pos')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 684,
   "id": "83f5a862",
   "metadata": {},
   "outputs": [],
   "source": [
    "pos_path = 'C:/Users/kojun/Desktop/ML/deep learning/文字/LMDB/aclImdb_v1/aclImdb/test/pos/'\n",
    "neg_path = 'C:/Users/kojun/Desktop/ML/deep learning/文字/LMDB/aclImdb_v1/aclImdb/test/neg/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 685,
   "id": "fc811a05",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████| 12500/12500 [00:00<00:00, 29584.29it/s]\n"
     ]
    }
   ],
   "source": [
    "test_one=[]\n",
    "test_pos_score = []\n",
    "test_pos_content = []\n",
    "for i in tqdm(range(len(test_pos_list))):\n",
    "    with open(os.path.join(pos_path,test_pos_list[i]),'r',encoding=\"utf-8\") as f:\n",
    "        test_pos_score.append(int(test_pos_list[i].split('.')[0].split('_')[1]))\n",
    "        test_pos_content.append(f.read())\n",
    "        test_one.append('pos')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 686,
   "id": "4ea88d66",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|█████████████████████████████████████████████████████████████████████████| 12500/12500 [00:00<00:00, 26472.79it/s]\n"
     ]
    }
   ],
   "source": [
    "test_zero=[]\n",
    "test_neg_score = []\n",
    "test_neg_content = []\n",
    "for i in tqdm(range(len(test_neg_list))):\n",
    "    with open(os.path.join(neg_path,test_neg_list[i]),'r',encoding=\"utf-8\") as f:\n",
    "        test_neg_score.append(int(test_neg_list[i].split('.')[0].split('_')[1]))\n",
    "        test_neg_content.append(f.read())\n",
    "        test_zero.append('neg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 687,
   "id": "112b6356",
   "metadata": {},
   "outputs": [],
   "source": [
    "test_call_content = test_pos_content+test_neg_content\n",
    "test_call_score = test_pos_score+test_neg_score\n",
    "test_call_one_zero = test_one+test_zero"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 688,
   "id": "dc3cb0bb",
   "metadata": {},
   "outputs": [],
   "source": [
    "test = pd.DataFrame()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 689,
   "id": "cbba3208",
   "metadata": {},
   "outputs": [],
   "source": [
    "test['Contain'] = test_call_content\n",
    "test['score'] = test_call_score\n",
    "test['ps_ne'] = test_call_one_zero"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 690,
   "id": "ef370739",
   "metadata": {},
   "outputs": [],
   "source": [
    "test.loc[test['ps_ne']=='pos','ps_ne'] = 1\n",
    "test.loc[test['ps_ne']=='neg','ps_ne'] = 0\n",
    "test['ps_ne'] = test['ps_ne'].astype('float16')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 691,
   "id": "84f67d25",
   "metadata": {},
   "outputs": [],
   "source": [
    "test['Contain'] = test.apply(lambda row:text_cleaning(row['Contain']),axis=1)\n",
    "test_1 = test.apply(lambda row:average_verc(row['Contain']),axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 692,
   "id": "8f5b5e0a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0        [0.0377221, -0.11066994, -0.0011112182, -0.044...\n",
       "1        [0.017637972, -0.030959306, 0.048518356, -0.00...\n",
       "2        [0.06898828, -0.04384847, 0.009476686, -0.0141...\n",
       "3        [0.06971887, -0.044306874, 0.02073427, 0.01709...\n",
       "4        [0.07801717, -0.07226931, -0.0037782707, -0.01...\n",
       "                               ...                        \n",
       "24995    [0.020210724, -0.10508037, 0.010371659, 0.0123...\n",
       "24996    [0.058934994, -0.093938656, 0.02708537, 0.0130...\n",
       "24997    [0.04310523, -0.031486597, 0.0023403151, -0.00...\n",
       "24998    [0.034909923, -0.056259338, -0.009873354, 0.01...\n",
       "24999    [0.08066439, -0.05083632, 0.006046505, -0.0047...\n",
       "Length: 25000, dtype: object"
      ]
     },
     "execution_count": 692,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 693,
   "id": "224e7990",
   "metadata": {},
   "outputs": [],
   "source": [
    "g= np.empty(shape=(25000,vector_size))\n",
    "for i in range(len(test)):\n",
    "    g[i] = test_1[i]\n",
    "g=pd.DataFrame(g)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 694,
   "id": "8ae661f0",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "782/782 [==============================] - 1s 1ms/step\n"
     ]
    }
   ],
   "source": [
    "y_pred = DNN_model.predict(g)\n",
    "y_pred = np.argmax(y_pred,axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 695,
   "id": "bb1211e0",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.86156"
      ]
     },
     "execution_count": 695,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "accuracy_score(test['ps_ne'],y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 696,
   "id": "9ca3a2ad",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[10830,  1670],\n",
       "       [ 1791, 10709]], dtype=int64)"
      ]
     },
     "execution_count": 696,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "confusion_matrix(test['ps_ne'],y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e2a24d21",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred=dnn_modedl.predict(test_w2v_x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4975eaf3",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "b3310e03",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.7"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}