让我们来看一个经典的儿分类问题：垃圾邮件过滤
    问题描述：因为我们用到的数据集是来自UCI机器学习仓库中的垃圾信息数据集，所以我们要分类的是垃圾短信和非垃圾短信😝，数据集我们可以从https://archive.ics.uci.edu/ml/datasets/sms+spam+collection下载。下面让我们来看一下代码。

    2. import numpy as np
    3. import pandas as pd
    5. from sklearn.feature_extraction.text import TfidfVectorizer
    6. from sklearn.model_selection import train_test_split, cross_val_score
    7. from sklearn.linear_model import LogisticRegression
    8. from sklearn.metrics import classification_report, f1_score
    11. def main():
    12.     df = pd.read_csv('./smsspamcollection/SMSSpamCollection', sep='\t', names=['label', 'content'])
    13.     print('垃圾短信数据量：{}条'.format(df[df['label'] == 'spam'].__len__()))
    14.     print('非垃圾短信数据集：{}条'.format(df[df['label'] == 'ham'].__len__()))
    15.     # 垃圾短信数据量：747条
    16.     # 非垃圾短信数据集：4825条
    18.     X_train_text, X_test_text, y_train, y_test = train_test_split(df['content'].values, df['label'].values,
    19.                                                                   test_size=0.25, shuffle=True, random_state=2020)
    20.     # 向量化
    21.     tfidf = TfidfVectorizer()
    22.     X_train = tfidf.fit_transform(X_train_text)
    23.     X_test = tfidf.transform(X_test_text)
    25.     logist_model = LogisticRegression()
    27.     logist_model.fit(X_train, y_train)
    29.     predictions = logist_model.predict(X_test)
    31.     for i, prediction in enumerate(predictions[:1]):
    32.         print('预测结果为：{}，短信内容为：{}'.format(prediction, X_test_text[i]))
    34.     # 预测结果为：ham，短信内容为：Hmm...my uncle just informed me that he's paying the school directly. So pls buy food.
    36.     # test label 向量化
    37.     y2int = dict(zip(logist_model.classes_, range(np.unique(y_test).shape[0])))
    38.     y_test = np.apply_along_axis(lambda x: y2int[x[0]], arr=y_test.reshape(-1, 1), axis=-1)
    40.     accuary = cross_val_score(logist_model, X_train, y_train, cv=5, scoring='accuracy')      # accuracy分数5折交叉验证
    41.     # train label 向量化
    42.     y_train = np.apply_along_axis(lambda x: y2int[x[0]], arr=y_train.reshape(-1, 1), axis=-1)
    44.     f1 = cross_val_score(logist_model, X_train, y_train, cv=5, scoring='f1')                # f1分数5折交叉验证
    45.     recall_score = cross_val_score(logist_model, X_train, y_train, cv=5, scoring='recall')        # recall分数5折交叉验证
    46.     precision_score = cross_val_score(logist_model, X_train, y_train, cv=5, scoring='precision')  # precision分数5折交叉验证
    48.     print(f'accuary得分：{accuary}\n'
    49.           f'f1_score得分：{f1}\n'
    50.           f'recall_score得分：{recall_score}\n'
    51.           f'precision_score得分：{precision_score}\n')
    52.     print(f'accuary5折交叉验证平均得分：{np.mean(accuary)}\n'
    53.           f'f1_score5折交叉验证平均得分：{np.mean(f1)}\n'
    54.           f'recall_score5折交叉验证平均得分：{np.mean(recall_score)}\n'
    55.           f'precision_score5折交叉验证平均得分：{np.mean(precision_score)}\n')
    56.     print(classification_report(y_test, np.apply_along_axis(lambda x: y2int[x[0]], arr=logist_model.predict(X_test).reshape(-1, 1), axis=-1)))
    58.     """
    59.         accuary得分：[0.965311   0.95095694 0.95574163 0.94497608 0.95688623]
    60.         f1_score得分：[0.85128205 0.77837838 0.8042328  0.74725275 0.80851064]
    61.         recall_score得分：[0.74107143 0.64285714 0.67857143 0.60714286 0.67857143]
    62.         precision_score得分：[1.         0.98630137 0.98701299 0.97142857 1.        ]
    64.         accuary5折交叉验证平均得分：0.9547743746955849
    65.         f1_score5折交叉验证平均得分：0.7979313238887707
    66.         recall_score5折交叉验证平均得分：0.6696428571428571
    67.         precision_score5折交叉验证平均得分：0.9889485856609145
    69.                   precision    recall  f1-score   support
    71.                0       0.97      1.00      0.98      1206
    72.                1       0.99      0.79      0.88       187
    74.         accuracy                           0.97      1393
    75.        macro avg       0.98      0.90      0.93      1393
    76.     weighted avg       0.97      0.97      0.97      1393
    77.     """
    80. if __name__ == '__main__':
    81.     main()