3-TextDoe-LSTM.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "source": [
        "# Resoure Preparation"
      ],
      "metadata": {
        "id": "s-Tb4wpOOZ-F"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Drive Mount"
      ],
      "metadata": {
        "id": "lyg54VvgOkdM"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ysNqS8IY0Idc",
        "outputId": "a0d38cca-fe9d-484a-c558-5f175428a2f4"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ],
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive',force_remount=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Setup Library"
      ],
      "metadata": {
        "id": "p2l9CALjOoOa"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "A4nXVQ_jyZbg",
        "outputId": "3b5f545a-8c2e-4f7e-8183-30f058d2fa72"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "\u001b[K     |████████████████████████████████| 4.6 MB 5.1 MB/s \n",
            "\u001b[K     |████████████████████████████████| 511.7 MB 6.4 kB/s \n",
            "\u001b[K     |████████████████████████████████| 438 kB 65.8 MB/s \n",
            "\u001b[K     |████████████████████████████████| 1.6 MB 52.8 MB/s \n",
            "\u001b[K     |████████████████████████████████| 5.8 MB 25.3 MB/s \n",
            "\u001b[K     |████████████████████████████████| 11.5 MB 937 kB/s \n",
            "\u001b[K     |████████████████████████████████| 965 kB 4.8 MB/s \n",
            "\u001b[?25h"
          ]
        }
      ],
      "source": [
        "!pip -q install tensorflow_text\n",
        "!pip -q install pythainlp \n",
        "!pip -q install python-crfsuite"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "y6oWkgLVpADc"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "\n",
        "from pythainlp import word_tokenize\n",
        "import re\n",
        "\n",
        "import gdown\n",
        "from google.colab import drive"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FzxV2lLFysBP",
        "outputId": "7700011e-e140-4dc7-836f-a1cb961a72c7"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?export=download&id=1ShyXkcRa-Yl2xDtdUqebBmwBEC_AhqH1\n",
            "To: /content/TNC_DataSet_5000_V1.zip\n",
            "100%|██████████| 46.9M/46.9M [00:00<00:00, 205MB/s]\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Archive:  TNC_DataSet_5000_V1.zip\n",
            "  inflating: dataset_info.txt        \n",
            "  inflating: TNC_AllDomain_5000.csv  \n",
            "  inflating: TNC_TestSet_5000.csv    \n",
            "  inflating: TNC_TrainSet_5000.csv   \n",
            "  inflating: TNC_ValidationSet_5000.csv  \n"
          ]
        }
      ],
      "source": [
        "driveURL = \"https://drive.google.com/file/d/1ShyXkcRa-Yl2xDtdUqebBmwBEC_AhqH1/view?usp=sharing\"\n",
        "downloadURL = 'https://drive.google.com/uc?export=download&id='+driveURL.split('/')[-2]\n",
        "gdown.download(downloadURL)\n",
        "!unzip TNC_DataSet_5000_V1.zip"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Data Preprocessing"
      ],
      "metadata": {
        "id": "iDXel_gTPGcx"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "p_SZjvvEy8l_"
      },
      "outputs": [],
      "source": [
        "def replace_url(text):\n",
        "    URL_PATTERN = r\"\"\"(?i)\\b((?:https?:(?:/{1,3}|[a-z0-9%])|[a-z0-9.\\-]+[.](?:com|net|org|edu|gov|mil|aero|asia|biz|cat|coop|info|int|jobs|mobi|museum|name|post|pro|tel|travel|xxx|ac|ad|ae|af|ag|ai|al|am|an|ao|aq|ar|as|at|au|aw|ax|az|ba|bb|bd|be|bf|bg|bh|bi|bj|bm|bn|bo|br|bs|bt|bv|bw|by|bz|ca|cc|cd|cf|cg|ch|ci|ck|cl|cm|cn|co|cr|cs|cu|cv|cx|cy|cz|dd|de|dj|dk|dm|do|dz|ec|ee|eg|eh|er|es|et|eu|fi|fj|fk|fm|fo|fr|ga|gb|gd|ge|gf|gg|gh|gi|gl|gm|gn|gp|gq|gr|gs|gt|gu|gw|gy|hk|hm|hn|hr|ht|hu|id|ie|il|im|in|io|iq|ir|is|it|je|jm|jo|jp|ke|kg|kh|ki|km|kn|kp|kr|kw|ky|kz|la|lb|lc|li|lk|lr|ls|lt|lu|lv|ly|ma|mc|md|me|mg|mh|mk|ml|mm|mn|mo|mp|mq|mr|ms|mt|mu|mv|mw|mx|my|mz|na|nc|ne|nf|ng|ni|nl|no|np|nr|nu|nz|om|pa|pe|pf|pg|ph|pk|pl|pm|pn|pr|ps|pt|pw|py|qa|re|ro|rs|ru|rw|sa|sb|sc|sd|se|sg|sh|si|sj|Ja|sk|sl|sm|sn|so|sr|ss|st|su|sv|sx|sy|sz|tc|td|tf|tg|th|tj|tk|tl|tm|tn|to|tp|tr|tt|tv|tw|tz|ua|ug|uk|us|uy|uz|va|vc|ve|vg|vi|vn|vu|wf|ws|ye|yt|yu|za|zm|zw)/)(?:[^\\s()<>{}\\[\\]]+|\\([^\\s()]*?\\([^\\s()]+\\)[^\\s()]*?\\)|\\([^\\s]+?\\))+(?:\\([^\\s()]*?\\([^\\s()]+\\)[^\\s()]*?\\)|\\([^\\s]+?\\)|[^\\s`!()\\[\\]{};:'\".,<>?«»“”‘’])|(?:(?<!@)[a-z0-9]+(?:[.\\-][a-z0-9]+)*[.](?:com|net|org|edu|gov|mil|aero|asia|biz|cat|coop|info|int|jobs|mobi|museum|name|post|pro|tel|travel|xxx|ac|ad|ae|af|ag|ai|al|am|an|ao|aq|ar|as|at|au|aw|ax|az|ba|bb|bd|be|bf|bg|bh|bi|bj|bm|bn|bo|br|bs|bt|bv|bw|by|bz|ca|cc|cd|cf|cg|ch|ci|ck|cl|cm|cn|co|cr|cs|cu|cv|cx|cy|cz|dd|de|dj|dk|dm|do|dz|ec|ee|eg|eh|er|es|et|eu|fi|fj|fk|fm|fo|fr|ga|gb|gd|ge|gf|gg|gh|gi|gl|gm|gn|gp|gq|gr|gs|gt|gu|gw|gy|hk|hm|hn|hr|ht|hu|id|ie|il|im|in|io|iq|ir|is|it|je|jm|jo|jp|ke|kg|kh|ki|km|kn|kp|kr|kw|ky|kz|la|lb|lc|li|lk|lr|ls|lt|lu|lv|ly|ma|mc|md|me|mg|mh|mk|ml|mm|mn|mo|mp|mq|mr|ms|mt|mu|mv|mw|mx|my|mz|na|nc|ne|nf|ng|ni|nl|no|np|nr|nu|nz|om|pa|pe|pf|pg|ph|pk|pl|pm|pn|pr|ps|pt|pw|py|qa|re|ro|rs|ru|rw|sa|sb|sc|sd|se|sg|sh|si|sj|Ja|sk|sl|sm|sn|so|sr|ss|st|su|sv|sx|sy|sz|tc|td|tf|tg|th|tj|tk|tl|tm|tn|to|tp|tr|tt|tv|tw|tz|ua|ug|uk|us|uy|uz|va|vc|ve|vg|vi|vn|vu|wf|ws|ye|yt|yu|za|zm|zw)\\b/?(?!@)))\"\"\"\n",
        "    return re.sub(URL_PATTERN, 'xxurl', text)\n",
        "\n",
        "def replace_rep(text):\n",
        "    def _replace_rep(m):\n",
        "        c,cc = m.groups()\n",
        "        return f'{c}xxrep'\n",
        "    re_rep = re.compile(r'(\\S)(\\1{2,})')\n",
        "    return re_rep.sub(_replace_rep, text)\n",
        "\n",
        "def process_text(text):\n",
        "    #pre rules\n",
        "    res = text.lower().strip() # 'AAAA'\n",
        "    res = replace_url(res)\n",
        "    res = replace_rep(res) # 'แมวนอนกินปลา' => 'แมว นอน กิน ปลา'\n",
        "    \n",
        "    #tokenize\n",
        "    res = [word for word in word_tokenize(res) if word and not re.search(pattern=r\"\\s+\", string=word)]\n",
        "    return res\n",
        "    \n",
        "DOMAIN_DICT = {\n",
        "      0 : 'Imaginative',\n",
        "      1 : 'Natural & Pure Science',\n",
        "      2 : 'Applied Science',\n",
        "      3 : 'Social Science ',\n",
        "      4 : 'History',\n",
        "      5 : 'Commerce & Finance',\n",
        "      6 : 'Arts',\n",
        "      7 : 'Belief & Thought'\n",
        "  }"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "65vFUp5Cy-09"
      },
      "outputs": [],
      "source": [
        "train_df = pd.read_csv(\"TNC_TrainSet_5000.csv\")\n",
        "valid_df = pd.read_csv(\"TNC_ValidationSet_5000.csv\")\n",
        "test_df = pd.read_csv(\"TNC_TestSet_5000.csv\")\n",
        "all_df = pd.concat([train_df,valid_df,test_df]).reset_index(drop=True)\n",
        "\n",
        "#Data Preprocessing\n",
        "train_df['Processed'] = train_df.Texts.map(lambda x: '|'.join(process_text(x)))\n",
        "valid_df['Processed'] = valid_df.Texts.map(lambda x: '|'.join(process_text(x)))\n",
        "test_df['Processed'] = test_df.Texts.map(lambda x: '|'.join(process_text(x)))\n",
        "\n",
        "#Shuffle Row\n",
        "train_df = train_df.sample(frac=1,random_state=69).reset_index(drop=True)\n",
        "valid_df = valid_df.sample(frac=1,random_state=69).reset_index(drop=True)\n",
        "test_df = test_df.sample(frac=1,random_state=69).reset_index(drop=True)"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Train a Long Short-Term Moemory (LSTM) Model"
      ],
      "metadata": {
        "id": "7wObK6k6PJ5l"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZQbXaapXzK8S",
        "outputId": "5508efe5-2c9c-484d-de66-8bda7bf15cd7"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(['สังคีต', 'ลักษณ์', 'รอน', 'โด', 'ของ'], 57480)"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ],
      "source": [
        "import tensorflow as tf \n",
        "\n",
        "\n",
        "text_train = [' '.join(x.split('|')) for x in train_df['Processed'].values.tolist()]\n",
        "text_valid = [' '.join(x.split('|')) for x in valid_df['Processed'].values.tolist()]\n",
        "text_test = [' '.join(x.split('|')) for x in test_df['Processed'].values.tolist()]\n",
        "\n",
        "word_count = []\n",
        "for sent in text_train:\n",
        "  for w in sent.split():\n",
        "      word_count.append(w)\n",
        "\n",
        "\n",
        "VOCAB_SIZE = len(set(word_count))\n",
        "encoder = tf.keras.layers.TextVectorization(max_tokens=VOCAB_SIZE)\n",
        "encoder.adapt(text_train)\n",
        "\n",
        "vocab = np.array(encoder.get_vocabulary())\n",
        "\n",
        "word_count[:5], VOCAB_SIZE "
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Zdw-SeVyz6p3",
        "outputId": "297d7043-b492-4ad5-c69f-edbc38f404a0"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "['Imaginative',\n",
              " 'Natural & Pure Science',\n",
              " 'Applied Science',\n",
              " 'Social Science ',\n",
              " 'History',\n",
              " 'Commerce & Finance',\n",
              " 'Arts',\n",
              " 'Belief & Thought']"
            ]
          },
          "metadata": {},
          "execution_count": 8
        }
      ],
      "source": [
        "y_train = train_df['Category']\n",
        "y_valid = valid_df['Category']\n",
        "y_test = test_df['Category']\n",
        "y_class = list(DOMAIN_DICT.values())\n",
        "y_class"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wCw0YN2Y0Bjg",
        "outputId": "ee14f620-3af1-48c8-c5d1-6a722fae495e"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(25200, 8)\n"
          ]
        }
      ],
      "source": [
        "from sklearn import preprocessing\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "\n",
        "le = preprocessing.LabelEncoder()\n",
        "le.fit(y_class)\n",
        "y_train = le.transform(y_train)\n",
        "y_val = le.transform(y_valid)\n",
        "y_train = to_categorical(y_train)\n",
        "y_val = to_categorical(y_val)\n",
        "print(y_train.shape)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "5UHTSfPB0Hi4"
      },
      "outputs": [],
      "source": [
        "model = tf.keras.Sequential([\n",
        "    encoder,\n",
        "    tf.keras.layers.Embedding(\n",
        "        input_dim=len(encoder.get_vocabulary()),\n",
        "        output_dim=512,\n",
        "        # Use masking to handle the variable sequence lengths\n",
        "        mask_zero=True),\n",
        "    tf.keras.layers.Bidirectional(tf.keras.layers.LSTM(256)),\n",
        "    tf.keras.layers.Dense(128, activation='relu'),\n",
        "    tf.keras.layers.Dense(y_train.shape[1], activation='softmax')\n",
        "])\n",
        "\n",
        "model.compile(loss='categorical_crossentropy',\n",
        "              optimizer=tf.keras.optimizers.Adam(1e-4),\n",
        "              metrics=['accuracy'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1Meja55S4r2Z",
        "outputId": "efbf8344-4c00-47da-c52e-62609630ca24"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "201600"
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ],
      "source": [
        "y_train.size"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "Ifsy-Lqf3rSg"
      },
      "outputs": [],
      "source": [
        "EPOCHS = 5\n",
        "BATCH_SIZE = 250\n",
        "checkpoint_path = '/content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/checkpoint2/CP_{epoch:04d}.ckpt'\n",
        "\n",
        "STEPS_PER_EPOCH = y_train.size / BATCH_SIZE\n",
        "SAVE_PERIOD = 1\n",
        "\n",
        "# Create a callback that saves the model's weights every 2 epochs\n",
        "cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path, \n",
        "                                                 verbose=1, \n",
        "                                                 save_weights_only=True,\n",
        "                                                 save_best_only = True,\n",
        "                                                 #save_freq= int(SAVE_PERIOD * STEPS_PER_EPOCH)\n",
        "                                                )\n",
        "\n",
        "# Train the model with the new callback\n",
        "def train_model():\n",
        "  model.fit(np.array(text_train),\n",
        "            y_train, \n",
        "            validation_data=(np.array(text_valid), y_val),\n",
        "            epochs=EPOCHS,\n",
        "            batch_size = BATCH_SIZE,\n",
        "            callbacks = [cp_callback],\n",
        "            verbose=2)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "iIRGlx1Ab9Ki",
        "outputId": "e40443f1-2bde-4408-ed31-20ebbb079503"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.training.tracking.util.CheckpointLoadStatus at 0x7fd407c73bd0>"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ],
      "source": [
        "#Load Model \n",
        "model.load_weights('/content/drive/MyDrive/AIB/TNC_LSTM_5000/checkpoint2/CP-Lastest')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "oVaBiT5mq75L",
        "outputId": "e4fc2d47-37e0-4da6-a4f0-30ccd894d884"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/5\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 1.68156, saving model to /content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/checkpoint2/CP_0001.ckpt\n",
            "101/101 - 3728s - loss: 0.0512 - accuracy: 0.9898 - val_loss: 1.6816 - val_accuracy: 0.7050 - 3728s/epoch - 37s/step\n",
            "Epoch 2/5\n",
            "\n",
            "Epoch 2: val_loss did not improve from 1.68156\n",
            "101/101 - 3706s - loss: 0.0457 - accuracy: 0.9900 - val_loss: 1.8314 - val_accuracy: 0.7059 - 3706s/epoch - 37s/step\n",
            "Epoch 3/5\n",
            "\n",
            "Epoch 3: val_loss did not improve from 1.68156\n",
            "101/101 - 3682s - loss: 0.0374 - accuracy: 0.9923 - val_loss: 1.8083 - val_accuracy: 0.7020 - 3682s/epoch - 36s/step\n",
            "Epoch 4/5\n",
            "\n",
            "Epoch 4: val_loss did not improve from 1.68156\n",
            "101/101 - 3715s - loss: 0.0318 - accuracy: 0.9932 - val_loss: 1.7369 - val_accuracy: 0.6930 - 3715s/epoch - 37s/step\n",
            "Epoch 5/5\n",
            "\n",
            "Epoch 5: val_loss improved from 1.68156 to 1.40053, saving model to /content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/checkpoint2/CP_0005.ckpt\n",
            "101/101 - 3728s - loss: 0.0795 - accuracy: 0.9817 - val_loss: 1.4005 - val_accuracy: 0.7174 - 3728s/epoch - 37s/step\n"
          ]
        }
      ],
      "source": [
        "train_model()\n",
        "model.save_weights('/content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/checkpoint3/CP-Lastest')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Vs1GI0zD78qL",
        "outputId": "5326f4f0-e8f7-4a57-be1a-1dde8a202dd6"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "WARNING:absl:Found untraced functions such as lstm_cell_1_layer_call_fn, lstm_cell_1_layer_call_and_return_conditional_losses, lstm_cell_2_layer_call_fn, lstm_cell_2_layer_call_and_return_conditional_losses while saving (showing 4 of 4). These functions will not be directly callable after loading.\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "INFO:tensorflow:Assets written to: /content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/saved_model/my_model/assets\n"
          ]
        },
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "INFO:tensorflow:Assets written to: /content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/saved_model/my_model/assets\n"
          ]
        }
      ],
      "source": [
        "!mkdir -p saved_model\n",
        "model.save('/content/drive/MyDrive/Programs/AIB/TNC_LSTM_5000/saved_model/my_model')"
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "# Model Evaluation"
      ],
      "metadata": {
        "id": "nxmTwrbYPXJO"
      }
    },
    {
      "cell_type": "markdown",
      "source": [
        "### Evaluate on validation set "
      ],
      "metadata": {
        "id": "OCF9_aRePZuU"
      }
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QG6fEwCA8Pj7",
        "outputId": "da7d0bce-526c-40db-e235-c78e031f72b9"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "85/85 [==============================] - 130s 2s/step\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.66      0.56      0.61       337\n",
            "           1       0.65      0.37      0.48       337\n",
            "           2       0.55      0.78      0.64       337\n",
            "           3       0.82      0.93      0.87       337\n",
            "           4       0.95      0.98      0.97       337\n",
            "           5       0.94      0.86      0.90       337\n",
            "           6       0.61      0.68      0.64       337\n",
            "           7       0.74      0.72      0.73       337\n",
            "\n",
            "    accuracy                           0.74      2696\n",
            "   macro avg       0.74      0.74      0.73      2696\n",
            "weighted avg       0.74      0.74      0.73      2696\n",
            "\n"
          ]
        }
      ],
      "source": [
        "from sklearn.metrics import classification_report\n",
        "value = model.predict(np.array(text_valid))\n",
        "y_val_pred = np.argmax(value,axis=1)\n",
        "y_val_true = np.argmax(y_val,axis=1)\n",
        "print(classification_report(y_val_true, y_val_pred))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "UfTHVwS786HD",
        "outputId": "b103a59d-98b7-4f76-a0f0-7b91dea8d399"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "array([[190,  49,  23,  29,   0,   0,  24,  22],\n",
              "       [ 58, 126,  63,   5,   0,   3,  62,  20],\n",
              "       [  3,   5, 263,   0,   0,  11,  52,   3],\n",
              "       [  1,   1,   0, 314,   0,   0,   0,  21],\n",
              "       [  0,   0,   0,   1, 331,   0,   0,   5],\n",
              "       [  0,   0,  45,   0,   0, 289,   3,   0],\n",
              "       [ 34,   4,  56,   2,   0,   1, 228,  12],\n",
              "       [  3,   8,  29,  33,  16,   2,   4, 242]])"
            ]
          },
          "execution_count": 21,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "#confusion matrix\n",
        "import numpy\n",
        "import matplotlib\n",
        "import matplotlib.pyplot as plt\n",
        "from sklearn.metrics import confusion_matrix, ConfusionMatrixDisplay\n",
        "from matplotlib.pyplot import figure\n",
        "\n",
        "\n",
        "cm = confusion_matrix(y_val_true, y_val_pred, labels=[0,1,2,3,4,5,6,7])\n",
        "cm"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 470
        },
        "id": "BqqnFwptdtC6",
        "outputId": "74cb543c-3d11-468c-c08c-f86fd76666d3"
      },
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<Figure size 100000x300000 with 0 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stderr",
          "output_type": "stream",
          "text": [
            "WARNING:matplotlib.font_manager:findfont: Font family ['normal'] not found. Falling back to DejaVu Sans.\n"
          ]
        },
        {
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "disp = ConfusionMatrixDisplay(confusion_matrix=cm,\n",
        "                              display_labels=y_class)\n",
        "\n",
        "font = {'family' : 'normal',\n",
        "        'size'   : 14}\n",
        "\n",
        "matplotlib.rc('font', **font)\n",
        "\n",
        "figure(figsize=(100, 300), dpi=1000)\n",
        "disp.plot()\n",
        "plt.xticks(rotation='vertical')\n",
        "plt.show()"
      ]
    }
  ],
  "metadata": {
    "colab": {
      "collapsed_sections": [],
      "name": "AIB_TNC_LSTM_5000.ipynb",
      "provenance": [],
      "toc_visible": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}