ND_Bert_DCard_titles_classification.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "ND-Bert-DCard-titles-classification.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/yehfanju/hello-world/blob/master/ND_Bert_DCard_titles_classification.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "w3S_GXyzPhd-"
      },
      "source": [
        "# DCard 標題分類"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "yV4RvkPsZ9_K"
      },
      "source": [
        "## 環境設置"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DV3WdHuEFfEZ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "33b6ece1-17c0-4535-85c5-50a5c80ed802"
      },
      "source": [
        "!gdown --id 1BRhhF1TXI8Z2BsSRmv20_ChOFu1r3-QR\n",
        "!pip install -q -U transformers"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Downloading...\n",
            "From: https://drive.google.com/uc?id=1BRhhF1TXI8Z2BsSRmv20_ChOFu1r3-QR\n",
            "To: /content/dcard_titles.csv\n",
            "\r  0% 0.00/44.9k [00:00<?, ?B/s]\r100% 44.9k/44.9k [00:00<00:00, 40.1MB/s]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "NC8wuoKU7nGH"
      },
      "source": [
        "import torch\r\n",
        "import torch.optim as optim\r\n",
        "import torch.nn as nn\r\n",
        "from tqdm.auto import tqdm\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "from sklearn.preprocessing import LabelEncoder\r\n",
        "from transformers import BertTokenizer, BertForSequenceClassification\r\n",
        "from sklearn.metrics import accuracy_score, confusion_matrix\r\n",
        "import numpy as np"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "2xwwmyfsXcih"
      },
      "source": [
        "## 資料前處理"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 402
        },
        "id": "kDlMfHZ2ExCh",
        "outputId": "8c18fbfb-b02c-4af0-fe31-766f1cb92175"
      },
      "source": [
        "import pandas as pd\n",
        "label_enc = LabelEncoder()\n",
        "titles = pd.read_csv(\"dcard_titles.csv\")\n",
        "titles = titles.sample(frac=1, random_state=12345)\n",
        "titles[\"label\"] = label_enc.fit_transform(titles.forum)\n",
        "titles"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>title</th>\n",
              "      <th>forum</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>144</th>\n",
              "      <td>calvin klein的皮帶如何？？</td>\n",
              "      <td>dressup</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>833</th>\n",
              "      <td>#問題 #問題 哪裡可以買砂紙</td>\n",
              "      <td>stationery</td>\n",
              "      <td>8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>270</th>\n",
              "      <td>心裡的檻卡住了🥲</td>\n",
              "      <td>relationship</td>\n",
              "      <td>7</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>468</th>\n",
              "      <td>#免費占卜 #免費八字 幫你卜一卦 第23篇</td>\n",
              "      <td>tarot</td>\n",
              "      <td>9</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>962</th>\n",
              "      <td>#有雷 拆彈專家2 中性評價</td>\n",
              "      <td>movie</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>420</th>\n",
              "      <td>塔羅牌義交流（花影&amp;偉特）</td>\n",
              "      <td>tarot</td>\n",
              "      <td>9</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>129</th>\n",
              "      <td>靴子怎麼挑？今年流行的妹子感及膝靴腿粗能穿嗎？掌握三點精準挑靴👢</td>\n",
              "      <td>dressup</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>285</th>\n",
              "      <td>無法陪伴對方而產生的罪惡感</td>\n",
              "      <td>relationship</td>\n",
              "      <td>7</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>485</th>\n",
              "      <td>大眾占卜 一週運勢(01/03-01/09)</td>\n",
              "      <td>tarot</td>\n",
              "      <td>9</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>482</th>\n",
              "      <td>#大眾占卜 塔羅星座周報2021.1.4-2021.1.10 ( 更新)</td>\n",
              "      <td>tarot</td>\n",
              "      <td>9</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>1000 rows × 3 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "                                    title         forum  label\n",
              "144                   calvin klein的皮帶如何？？       dressup      1\n",
              "833                       #問題 #問題 哪裡可以買砂紙    stationery      8\n",
              "270                              心裡的檻卡住了🥲  relationship      7\n",
              "468                #免費占卜 #免費八字 幫你卜一卦 第23篇         tarot      9\n",
              "962                        #有雷 拆彈專家2 中性評價         movie      4\n",
              "..                                    ...           ...    ...\n",
              "420                         塔羅牌義交流（花影&偉特）         tarot      9\n",
              "129      靴子怎麼挑？今年流行的妹子感及膝靴腿粗能穿嗎？掌握三點精準挑靴👢       dressup      1\n",
              "285                         無法陪伴對方而產生的罪惡感  relationship      7\n",
              "485                大眾占卜 一週運勢(01/03-01/09)         tarot      9\n",
              "482  #大眾占卜 塔羅星座周報2021.1.4-2021.1.10 ( 更新)         tarot      9\n",
              "\n",
              "[1000 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 3
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QYdgRIuPaHvh"
      },
      "source": [
        "## 模型定義"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CZz3LGejE3cG",
        "outputId": "988e2fe2-1a4b-43bd-db09-42d2bd13e5f5"
      },
      "source": [
        "tokenizer = BertTokenizer.from_pretrained(\"bert-base-chinese\")\n",
        "model = BertForSequenceClassification.from_pretrained('bert-base-chinese', num_labels=len(label_enc.classes_))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Some weights of the model checkpoint at bert-base-chinese were not used when initializing BertForSequenceClassification: ['cls.predictions.bias', 'cls.predictions.transform.dense.weight', 'cls.predictions.transform.dense.bias', 'cls.predictions.decoder.weight', 'cls.seq_relationship.weight', 'cls.seq_relationship.bias', 'cls.predictions.transform.LayerNorm.weight', 'cls.predictions.transform.LayerNorm.bias']\n",
            "- This IS expected if you are initializing BertForSequenceClassification from the checkpoint of a model trained on another task or with another architecture (e.g. initializing a BertForSequenceClassification model from a BertForPreTraining model).\n",
            "- This IS NOT expected if you are initializing BertForSequenceClassification from the checkpoint of a model that you expect to be exactly identical (initializing a BertForSequenceClassification model from a BertForSequenceClassification model).\n",
            "Some weights of BertForSequenceClassification were not initialized from the model checkpoint at bert-base-chinese and are newly initialized: ['classifier.weight', 'classifier.bias']\n",
            "You should probably TRAIN this model on a down-stream task to be able to use it for predictions and inference.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-S2oddSnaNb-",
        "outputId": "365c5e83-0f7d-49b7-908b-6078c509f831"
      },
      "source": [
        "tokenizer([\"電影\", \"五月天\"], padding=True, return_tensors=\"pt\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "{'input_ids': tensor([[ 101, 7442, 2512,  102,    0],\n",
              "        [ 101,  758, 3299, 1921,  102]]), 'token_type_ids': tensor([[0, 0, 0, 0, 0],\n",
              "        [0, 0, 0, 0, 0]]), 'attention_mask': tensor([[1, 1, 1, 1, 0],\n",
              "        [1, 1, 1, 1, 1]])}"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XfpQ3ivwicmo"
      },
      "source": [
        "## 訓練迴圈"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ulybNNPELbpe",
        "outputId": "2edaf884-fdf2-46f4-c10b-48138665bc28"
      },
      "source": [
        "batch_size = 16\n",
        "n_epoch = 5\n",
        "torch.manual_seed(12345)\n",
        "\n",
        "train = titles.iloc[:900]\n",
        "test = titles.iloc[900:]\n",
        "\n",
        "device = torch.device(\"cuda\") if torch.cuda.is_available() else torch.device(\"cpu\")\n",
        "model.to(device)\n",
        "model.train()\n",
        "\n",
        "loss_fn = torch.nn.CrossEntropyLoss()\n",
        "optimizer = torch.optim.AdamW(model.parameters(), lr=5e-5)\n",
        "\n",
        "loss_train = []\n",
        "acc_train = []\n",
        "acc_test = []\n",
        "\n",
        "for epoch_i in range(n_epoch):\n",
        "  loss_train_batch = []\n",
        "  acc_train_batch = []\n",
        "\n",
        "  for idx in range(0, train.shape[0], batch_size):\n",
        "    optimizer.zero_grad()\n",
        "    batch = train.iloc[idx:idx+batch_size]\n",
        "    X = tokenizer(batch.title.tolist(), padding=True, return_tensors=\"pt\").to(device)\n",
        "    y = torch.tensor(batch.label.tolist(), dtype=torch.long).to(device)\n",
        "    out = model(**X)\n",
        "    loss = loss_fn(out.logits, y)\n",
        "    loss.backward()\n",
        "    optimizer.step()\n",
        "\n",
        "    preds = out.logits.argmax(axis=1).detach().cpu().numpy()\n",
        "    acc = accuracy_score(y.cpu().numpy(), preds)\n",
        "    loss_train_batch.append(loss.detach().cpu().item())\n",
        "    acc_train_batch.append(acc)\n",
        "  \n",
        "  ## validation\n",
        "  acc_test_batch = []\n",
        "  ytrue_test = []\n",
        "  ypred_test = []\n",
        "  model.eval()\n",
        "  for idx in range(0, test.shape[0], batch_size):\n",
        "    batch = test.iloc[idx:idx+batch_size]\n",
        "    X = tokenizer(batch.title.tolist(), padding=True, return_tensors=\"pt\").to(device)\n",
        "    y = torch.tensor(batch.label.tolist(), dtype=torch.long).to(device)\n",
        "    out = model(**X)\n",
        "\n",
        "    preds = out.logits.argmax(axis=1).detach().cpu().numpy()\n",
        "    acc = accuracy_score(y.cpu().numpy(), preds)\n",
        "    acc_test_batch.append(acc)\n",
        "    ytrue_test.extend(y.cpu().numpy())\n",
        "    ypred_test.extend(preds)\n",
        "  loss_train.append(np.mean(loss_train_batch))\n",
        "  acc_train.append(np.mean(acc_train_batch))\n",
        "  acc_test.append(np.mean(acc_test_batch))\n",
        "  print(\"[Epoch {}] Train loss: {:.4f}, \".format(epoch_i, loss_train[-1]), end='')\n",
        "  print(\"Train Acc: {:.4f}, \".format(acc_train[-1]), end='')\n",
        "  print(\"Test Acc: {:.4f}, \".format(acc_test[-1]))\n",
        "\n",
        "  "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "[Epoch 0] Train loss: 1.7040, Train Acc: 0.4781, Test Acc: 0.7857, \n",
            "[Epoch 1] Train loss: 0.6745, Train Acc: 0.7993, Test Acc: 0.7589, \n",
            "[Epoch 2] Train loss: 0.3209, Train Acc: 0.9057, Test Acc: 0.7768, \n",
            "[Epoch 3] Train loss: 0.1349, Train Acc: 0.9660, Test Acc: 0.8125, \n",
            "[Epoch 4] Train loss: 0.0555, Train Acc: 0.9890, Test Acc: 0.7857, \n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "JiS5p_LfbuAV"
      },
      "source": [
        "## 混淆矩陣"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "L0fYBlBlb1g8"
      },
      "source": [
        "Confusion matrix whose i-th row and j-th column entry indicates the number of samples with true label being i-th class and predicted label being j-th class."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 373
        },
        "id": "lld9A5KiRtJB",
        "outputId": "2672c633-38e0-4f9d-ba15-46bf77366eec"
      },
      "source": [
        "confmat = confusion_matrix(ytrue_test, ypred_test)\n",
        "conf = pd.DataFrame(confmat, index=label_enc.classes_, columns=label_enc.classes_)\n",
        "row_sum = conf.sum(0)\n",
        "row_sum.name = \"pred_sum\"\n",
        "conf = conf.append(row_sum)\n",
        "conf[\"label_sum\"] = conf.sum(axis=1)\n",
        "conf"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "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>boy</th>\n",
              "      <th>dressup</th>\n",
              "      <th>girl</th>\n",
              "      <th>makeup</th>\n",
              "      <th>movie</th>\n",
              "      <th>music</th>\n",
              "      <th>psychology</th>\n",
              "      <th>relationship</th>\n",
              "      <th>stationery</th>\n",
              "      <th>tarot</th>\n",
              "      <th>label_sum</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>boy</th>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>dressup</th>\n",
              "      <td>0</td>\n",
              "      <td>8</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>11</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>girl</th>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>9</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>makeup</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>13</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>15</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>movie</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>4</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>6</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>music</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>9</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>psychology</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>13</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>13</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>relationship</th>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>2</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>8</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>12</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>stationery</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "      <td>0</td>\n",
              "      <td>7</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>tarot</th>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>6</td>\n",
              "      <td>8</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>pred_sum</th>\n",
              "      <td>7</td>\n",
              "      <td>9</td>\n",
              "      <td>10</td>\n",
              "      <td>14</td>\n",
              "      <td>7</td>\n",
              "      <td>11</td>\n",
              "      <td>15</td>\n",
              "      <td>13</td>\n",
              "      <td>7</td>\n",
              "      <td>7</td>\n",
              "      <td>100</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "              boy  dressup  girl  ...  stationery  tarot  label_sum\n",
              "boy             5        0     1  ...           0      0          7\n",
              "dressup         0        8     1  ...           0      0         11\n",
              "girl            0        1     3  ...           0      0          9\n",
              "makeup          0        0     1  ...           0      0         15\n",
              "movie           0        0     1  ...           0      0          6\n",
              "music           1        0     1  ...           0      0         12\n",
              "psychology      0        0     0  ...           0      0         13\n",
              "relationship    1        0     2  ...           0      1         12\n",
              "stationery      0        0     0  ...           7      0          7\n",
              "tarot           0        0     0  ...           0      6          8\n",
              "pred_sum        7        9    10  ...           7      7        100\n",
              "\n",
              "[11 rows x 11 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "IGtz93wDakEL"
      },
      "source": [
        "## 預測/推論"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q8U7gsmyKeyC"
      },
      "source": [
        "def predict(text):\n",
        "  X = tokenizer([text], padding=True, return_tensors=\"pt\").to(device)\n",
        "  out = model(**X)\n",
        "\n",
        "  probs = out.logits\n",
        "  preds = probs.argmax(axis=1).detach().cpu().numpy()\n",
        "  probs = probs.softmax(axis=1).squeeze().detach().cpu().numpy()\n",
        "\n",
        "  plt.figure(facecolor=\"white\")\n",
        "  xs = np.arange(len(label_enc.classes_))\n",
        "  plt.bar(xs, height=probs)\n",
        "  plt.xticks(xs, labels=label_enc.classes_, rotation=45)\n",
        "  return label_enc.inverse_transform(preds)[0]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 304
        },
        "id": "L2JqmOvmPsap",
        "outputId": "e7fd1b0d-a613-43d2-a476-5150304cd2f3"
      },
      "source": [
        "prob = predict(\"靈魂急轉彎好看嗎\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 322
        },
        "id": "oNno7ct3edhA",
        "outputId": "248136bd-8c82-4674-b0d4-166ecb108e18"
      },
      "source": [
        "predict(\"所有感情問題都建議分手\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'relationship'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 10
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 322
        },
        "id": "BiwTo-_4elQb",
        "outputId": "b8affacc-ed76-4c7c-9988-35f50c7d990d"
      },
      "source": [
        "predict(\"五月天和蘇打綠\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'music'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 11
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "8EO-EtW9erGA"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}