{"id":1180621,"date":"2025-01-15T18:39:44","date_gmt":"2025-01-15T10:39:44","guid":{"rendered":""},"modified":"2025-01-15T18:40:04","modified_gmt":"2025-01-15T10:40:04","slug":"python-%e8%a1%a8%e6%a0%bc%e5%a6%82%e4%bd%95%e5%8e%bb%e9%99%a4%e5%bc%82%e5%b8%b8%e8%a1%8c","status":"publish","type":"post","link":"https:\/\/docs.pingcode.com\/ask\/1180621.html","title":{"rendered":"python \u8868\u683c\u5982\u4f55\u53bb\u9664\u5f02\u5e38\u884c"},"content":{"rendered":"

\"python<\/p>\n

\u5728 Python \u4e2d\u53bb\u9664\u8868\u683c\u4e2d\u7684\u5f02\u5e38\u884c\u53ef\u4ee5\u901a\u8fc7\u591a\u79cd\u65b9\u6cd5\u6765\u5b9e\u73b0\uff0c\u5e38\u7528\u7684\u65b9\u6cd5\u6709\u4f7f\u7528 pandas \u5e93\u3001\u6570\u636e\u6e05\u6d17\u548c\u8fc7\u6ee4\u3001\u7edf\u8ba1\u5206\u6790\u548c\u53ef\u89c6\u5316<\/strong>\u7b49\u3002\u4f7f\u7528 pandas \u5e93<\/strong>\u662f\u4e00\u79cd\u975e\u5e38\u9ad8\u6548\u7684\u65b9\u6cd5\uff0c\u56e0\u4e3a\u5b83\u63d0\u4f9b\u4e86\u5f3a\u5927\u7684\u6570\u636e\u64cd\u4f5c\u529f\u80fd\u3002\u4e0b\u9762\u6211\u4eec\u5c06\u8be6\u7ec6\u4ecb\u7ecd\u5982\u4f55\u4f7f\u7528 pandas \u5e93\u6765\u53bb\u9664\u5f02\u5e38\u884c\u3002<\/p>\n<\/p>\n

\u4e00\u3001\u4f7f\u7528 Pandas \u5e93<\/h3>\n<\/p>\n

Pandas \u662f\u4e00\u4e2a\u5f3a\u5927\u7684\u6570\u636e\u5904\u7406\u5e93\uff0c\u63d0\u4f9b\u4e86\u8bb8\u591a\u65b9\u4fbf\u7684\u6570\u636e\u64cd\u4f5c\u65b9\u6cd5\u3002\u5728\u5904\u7406\u5f02\u5e38\u884c\u65f6\uff0c\u901a\u5e38\u4f7f\u7528\u4ee5\u4e0b\u6b65\u9aa4\uff1a<\/p>\n<\/p>\n

    \n
  1. \u8bfb\u53d6\u6570\u636e<\/strong><\/li>\n
  2. \u8bc6\u522b\u5f02\u5e38\u503c<\/strong><\/li>\n
  3. \u53bb\u9664\u5f02\u5e38\u884c<\/strong><\/li>\n<\/ol>\n

    1.1 \u8bfb\u53d6\u6570\u636e<\/h4>\n<\/p>\n

    \u9996\u5148\uff0c\u6211\u4eec\u9700\u8981\u8bfb\u53d6\u6570\u636e\uff0c\u8fd9\u901a\u5e38\u662f\u4ece CSV \u6587\u4ef6\u6216 Excel \u6587\u4ef6\u4e2d\u8bfb\u53d6\u3002Pandas \u63d0\u4f9b\u4e86 read_csv<\/code> \u548c read_excel<\/code> \u65b9\u6cd5\u6765\u8bfb\u53d6\u6570\u636e\u3002<\/p>\n<\/p>\n

    import pandas as pd<\/p>\n
    \u8bfb\u53d6 CSV \u6587\u4ef6<\/strong><\/h2>\n
    df = pd.read_csv('data.csv')<\/p>\n
    \u8bfb\u53d6 Excel \u6587\u4ef6<\/strong><\/h2>\n
    df = pd.read_excel('data.xlsx')<\/strong><\/h2>\n
    <\/code><\/pre>\n<\/p>\n
    1.2 \u8bc6\u522b\u5f02\u5e38\u503c<\/h4>\n<\/p>\n
    \u8bc6\u522b\u5f02\u5e38\u503c\u662f\u53bb\u9664\u5f02\u5e38\u884c\u7684\u5173\u952e\u6b65\u9aa4\u3002\u901a\u5e38\uff0c\u53ef\u4ee5\u4f7f\u7528\u4ee5\u4e0b\u51e0\u79cd\u65b9\u6cd5\u6765\u8bc6\u522b\u5f02\u5e38\u503c\uff1a<\/p>\n<\/p>\n
      \n
    • \u7edf\u8ba1\u65b9\u6cd5<\/strong>\uff1a\u5982\u5747\u503c\u548c\u6807\u51c6\u5dee<\/li>\n
    • \u7bb1\u7ebf\u56fe<\/strong>\uff1a\u4f7f\u7528\u56db\u5206\u4f4d\u6570<\/li>\n
    • Z-score<\/strong>\uff1a\u6807\u51c6\u5316\u5206\u6570<\/li>\n<\/ul>\n
      1.2.1 \u4f7f\u7528\u7edf\u8ba1\u65b9\u6cd5<\/h5>\n<\/p>\n
      \u901a\u8fc7\u8ba1\u7b97\u5747\u503c\u548c\u6807\u51c6\u5dee\uff0c\u53ef\u4ee5\u8bc6\u522b\u51fa\u4e0e\u5747\u503c\u5dee\u8ddd\u8f83\u5927\u7684\u5f02\u5e38\u503c\u3002<\/p>\n<\/p>\n
      mean = df['column_name'].mean()<\/p>\n
      std = df['column_name'].std()<\/p>\n
      threshold = 3  # \u901a\u5e38\u53d63\u500d\u6807\u51c6\u5dee<\/p>\n
      \u8bc6\u522b\u5f02\u5e38\u503c<\/strong><\/h2>\n
      df['is_outlier'] = abs(df['column_name'] - mean) > threshold * std<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      1.2.2 \u4f7f\u7528\u7bb1\u7ebf\u56fe<\/h5>\n<\/p>\n
      \u7bb1\u7ebf\u56fe\u662f\u4e00\u79cd\u6709\u6548\u7684\u5f02\u5e38\u503c\u68c0\u6d4b\u65b9\u6cd5\uff0c\u4f7f\u7528\u56db\u5206\u4f4d\u6570\u6765\u8bc6\u522b\u5f02\u5e38\u503c\u3002<\/p>\n<\/p>\n
      Q1 = df['column_name'].quantile(0.25)<\/p>\n
      Q3 = df['column_name'].quantile(0.75)<\/p>\n
      IQR = Q3 - Q1  # \u56db\u5206\u4f4d\u8ddd<\/p>\n
      \u8bc6\u522b\u5f02\u5e38\u503c<\/strong><\/h2>\n
      df['is_outlier'] = (df['column_name'] < (Q1 - 1.5 * IQR)) | (df['column_name'] > (Q3 + 1.5 * IQR))<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      1.2.3 \u4f7f\u7528 Z-score<\/h5>\n<\/p>\n
      Z-score \u662f\u4e00\u79cd\u6807\u51c6\u5316\u5206\u6570\uff0c\u53ef\u4ee5\u7528\u6765\u8bc6\u522b\u4e0e\u5747\u503c\u5dee\u8ddd\u8f83\u5927\u7684\u5f02\u5e38\u503c\u3002<\/p>\n<\/p>\n
      from scipy.stats import zscore<\/p>\n
      \u8ba1\u7b97 Z-score<\/strong><\/h2>\n
      df['z_score'] = zscore(df['column_name'])<\/p>\n
      \u8bc6\u522b\u5f02\u5e38\u503c<\/strong><\/h2>\n
      df['is_outlier'] = abs(df['z_score']) > 3<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      1.3 \u53bb\u9664\u5f02\u5e38\u884c<\/h4>\n<\/p>\n
      \u8bc6\u522b\u51fa\u5f02\u5e38\u503c\u540e\uff0c\u53ef\u4ee5\u4f7f\u7528\u5e03\u5c14\u7d22\u5f15\u6765\u53bb\u9664\u5f02\u5e38\u884c\u3002<\/p>\n<\/p>\n
      # \u53bb\u9664\u5f02\u5e38\u884c<\/p>\n
      df_cleaned = df[~df['is_outlier']]<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      \u4e8c\u3001\u6570\u636e\u6e05\u6d17\u548c\u8fc7\u6ee4<\/h3>\n<\/p>\n
      \u6570\u636e\u6e05\u6d17\u662f\u53bb\u9664\u5f02\u5e38\u503c\u7684\u91cd\u8981\u6b65\u9aa4\u3002\u901a\u8fc7\u5bf9\u6570\u636e\u8fdb\u884c\u6e05\u6d17\uff0c\u53ef\u4ee5\u63d0\u9ad8\u6570\u636e\u7684\u8d28\u91cf\u548c\u53ef\u9760\u6027\u3002<\/p>\n<\/p>\n
      2.1 \u6570\u636e\u6e05\u6d17<\/h4>\n<\/p>\n
      \u6570\u636e\u6e05\u6d17\u5305\u62ec\u5904\u7406\u7f3a\u5931\u503c\u3001\u91cd\u590d\u503c\u548c\u4e0d\u4e00\u81f4\u7684\u6570\u636e\u3002<\/p>\n<\/p>\n
      # \u5904\u7406\u7f3a\u5931\u503c<\/p>\n
      df = df.dropna()  # \u5220\u9664\u7f3a\u5931\u503c<\/p>\n
      df = df.fillna(method='ffill')  # \u586b\u5145\u7f3a\u5931\u503c<\/strong><\/h2>\n
      \u5904\u7406\u91cd\u590d\u503c<\/strong><\/h2>\n
      df = df.drop_duplicates()<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      2.2 \u6570\u636e\u8fc7\u6ee4<\/h4>\n<\/p>\n
      \u6570\u636e\u8fc7\u6ee4\u662f\u901a\u8fc7\u6761\u4ef6\u9009\u62e9\u6765\u53bb\u9664\u4e0d\u7b26\u5408\u8981\u6c42\u7684\u6570\u636e\u3002<\/p>\n<\/p>\n
      # \u8fc7\u6ee4\u6570\u636e<\/p>\n
      df_filtered = df[df['column_name'] > threshold]<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      \u4e09\u3001\u7edf\u8ba1\u5206\u6790\u548c\u53ef\u89c6\u5316<\/h3>\n<\/p>\n
      \u901a\u8fc7\u7edf\u8ba1\u5206\u6790\u548c\u53ef\u89c6\u5316\uff0c\u53ef\u4ee5\u66f4\u76f4\u89c2\u5730\u8bc6\u522b\u5f02\u5e38\u503c\u3002<\/p>\n<\/p>\n
      3.1 \u7edf\u8ba1\u5206\u6790<\/h4>\n<\/p>\n
      \u7edf\u8ba1\u5206\u6790\u53ef\u4ee5\u5e2e\u52a9\u8bc6\u522b\u6570\u636e\u4e2d\u7684\u5f02\u5e38\u6a21\u5f0f\u3002<\/p>\n<\/p>\n
      # \u8ba1\u7b97\u63cf\u8ff0\u6027\u7edf\u8ba1\u91cf<\/p>\n
      desc_stats = df.describe()<\/p>\n
      print(desc_stats)<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      3.2 \u6570\u636e\u53ef\u89c6\u5316<\/h4>\n<\/p>\n
      \u6570\u636e\u53ef\u89c6\u5316\u53ef\u4ee5\u76f4\u89c2\u5730\u5c55\u793a\u6570\u636e\u4e2d\u7684\u5f02\u5e38\u503c\u3002<\/p>\n<\/p>\n
      import matplotlib.pyplot as plt<\/p>\n
      import seaborn as sns<\/p>\n
      \u7ed8\u5236\u7bb1\u7ebf\u56fe<\/strong><\/h2>\n
      sns.boxplot(x=df['column_name'])<\/p>\n
      plt.show()<\/p>\n
      \u7ed8\u5236\u6563\u70b9\u56fe<\/strong><\/h2>\n
      sns.scatterplot(x=df['column_x'], y=df['column_y'])<\/p>\n
      plt.show()<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      \u56db\u3001\u5f02\u5e38\u503c\u5904\u7406\u65b9\u6cd5<\/h3>\n<\/p>\n
      \u8bc6\u522b\u51fa\u5f02\u5e38\u503c\u540e\uff0c\u53ef\u4ee5\u9009\u62e9\u4e0d\u540c\u7684\u5904\u7406\u65b9\u6cd5\uff0c\u5982\u5220\u9664\u3001\u66ff\u6362\u6216\u8c03\u6574\u3002<\/p>\n<\/p>\n
      4.1 \u5220\u9664\u5f02\u5e38\u503c<\/h4>\n<\/p>\n
      \u5220\u9664\u5f02\u5e38\u503c\u662f\u4e00\u79cd\u76f4\u63a5\u7684\u65b9\u6cd5\uff0c\u4f46\u53ef\u80fd\u4f1a\u5bfc\u81f4\u6570\u636e\u91cf\u51cf\u5c11\u3002<\/p>\n<\/p>\n
      # \u5220\u9664\u5f02\u5e38\u503c<\/p>\n
      df = df[~df['is_outlier']]<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      4.2 \u66ff\u6362\u5f02\u5e38\u503c<\/h4>\n<\/p>\n
      \u66ff\u6362\u5f02\u5e38\u503c\u53ef\u4ee5\u4f7f\u7528\u5747\u503c\u3001\u4e2d\u4f4d\u6570\u6216\u5176\u4ed6\u5408\u7406\u7684\u503c\u3002<\/p>\n<\/p>\n
      # \u66ff\u6362\u5f02\u5e38\u503c<\/p>\n
      df.loc[df['is_outlier'], 'column_name'] = df['column_name'].median()<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      4.3 \u8c03\u6574\u5f02\u5e38\u503c<\/h4>\n<\/p>\n
      \u8c03\u6574\u5f02\u5e38\u503c\u662f\u901a\u8fc7\u9650\u5236\u5176\u8303\u56f4\u6765\u5904\u7406\u5f02\u5e38\u503c\u3002<\/p>\n<\/p>\n
      # \u8c03\u6574\u5f02\u5e38\u503c<\/p>\n
      df['column_name'] = df['column_name'].clip(lower=Q1 - 1.5 * IQR, upper=Q3 + 1.5 * IQR)<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      \u4e94\u3001\u7efc\u5408\u5b9e\u4f8b<\/h3>\n<\/p>\n
      \u4e0b\u9762\u662f\u4e00\u4e2a\u7efc\u5408\u5b9e\u4f8b\uff0c\u5c55\u793a\u4e86\u5982\u4f55\u4f7f\u7528 Pandas \u5e93\u6765\u53bb\u9664\u5f02\u5e38\u884c\u3002<\/p>\n<\/p>\n
      import pandas as pd<\/p>\n
      from scipy.stats import zscore<\/p>\n
      import matplotlib.pyplot as plt<\/p>\n
      import seaborn as sns<\/p>\n
      \u8bfb\u53d6\u6570\u636e<\/strong><\/h2>\n
      df = pd.read_csv('data.csv')<\/p>\n
      \u8ba1\u7b97 Z-score<\/strong><\/h2>\n
      df['z_score'] = zscore(df['column_name'])<\/p>\n
      \u8bc6\u522b\u5f02\u5e38\u503c<\/strong><\/h2>\n
      df['is_outlier'] = abs(df['z_score']) > 3<\/p>\n
      \u53bb\u9664\u5f02\u5e38\u884c<\/strong><\/h2>\n
      df_cleaned = df[~df['is_outlier']]<\/p>\n
      \u7ed8\u5236\u7bb1\u7ebf\u56fe<\/strong><\/h2>\n
      sns.boxplot(x=df_cleaned['column_name'])<\/p>\n
      plt.show()<\/p>\n
      \u7ed8\u5236\u6563\u70b9\u56fe<\/strong><\/h2>\n
      sns.scatterplot(x=df_cleaned['column_x'], y=df_cleaned['column_y'])<\/p>\n
      plt.show()<\/p>\n
      \u6253\u5370\u6e05\u6d17\u540e\u7684\u6570\u636e<\/strong><\/h2>\n
      print(df_cleaned)<\/p>\n
      <\/code><\/pre>\n<\/p>\n
      \u901a\u8fc7\u4ee5\u4e0a\u6b65\u9aa4\uff0c\u6211\u4eec\u53ef\u4ee5\u9ad8\u6548\u5730\u53bb\u9664\u8868\u683c\u4e2d\u7684\u5f02\u5e38\u884c\uff0c\u63d0\u9ad8\u6570\u636e\u7684\u8d28\u91cf\u548c\u53ef\u9760\u6027\u3002\u5e0c\u671b\u8fd9\u4e9b\u65b9\u6cd5\u5bf9\u4f60\u6709\u6240\u5e2e\u52a9\uff01<\/p>\n<\/p>\n
      \u76f8\u5173\u95ee\u7b54FAQs\uff1a<\/strong><\/h2>\n
       \u5982\u4f55\u8bc6\u522b\u8868\u683c\u4e2d\u7684\u5f02\u5e38\u884c\uff1f<\/strong>
      \u5728\u5904\u7406\u8868\u683c\u6570\u636e\u65f6\uff0c\u8bc6\u522b\u5f02\u5e38\u884c\u662f\u5173\u952e\u7684\u7b2c\u4e00\u6b65\u3002\u53ef\u4ee5\u901a\u8fc7\u591a\u79cd\u65b9\u5f0f\u6765\u53d1\u73b0\u8fd9\u4e9b\u5f02\u5e38\uff0c\u5305\u62ec\u68c0\u67e5\u7f3a\u5931\u503c\u3001\u8d85\u51fa\u9884\u671f\u8303\u56f4\u7684\u6570\u503c\u6216\u683c\u5f0f\u4e0d\u4e00\u81f4\u7684\u6761\u76ee\u3002\u4f7f\u7528Python\u4e2d\u7684Pandas\u5e93\uff0c\u53ef\u4ee5\u5229\u7528isnull()<\/code>\u548cdescribe()<\/code>\u65b9\u6cd5\u6765\u5e2e\u52a9\u8bc6\u522b\u6570\u636e\u4e2d\u7684\u5f02\u5e38\u60c5\u51b5\u3002<\/p>\n
      \u4f7f\u7528\u54ea\u79cd\u65b9\u6cd5\u53ef\u4ee5\u6709\u6548\u53bb\u9664\u5f02\u5e38\u884c\uff1f<\/strong>
      \u53bb\u9664\u5f02\u5e38\u884c\u7684\u65b9\u6cd5\u6709\u5f88\u591a\uff0c\u6700\u5e38\u7528\u7684\u5305\u62ec\u57fa\u4e8e\u6761\u4ef6\u7b5b\u9009\u548c\u4f7f\u7528\u7edf\u8ba1\u65b9\u6cd5\u3002\u5bf9\u4e8e\u7b80\u5355\u7684\u6761\u4ef6\uff0c\u53ef\u4ee5\u4f7f\u7528\u5e03\u5c14\u7d22\u5f15\u76f4\u63a5\u8fc7\u6ee4\u6570\u636e\u3002\u5bf9\u4e8e\u590d\u6742\u7684\u5f02\u5e38\u68c0\u6d4b\uff0c\u53ef\u4ee5\u8003\u8651\u4f7f\u7528Z-score\u6216IQR\uff08\u56db\u5206\u4f4d\u8ddd\uff09\u65b9\u6cd5\uff0c\u7ed3\u5408Pandas\u7684drop()<\/code>\u51fd\u6570\u6765\u5220\u9664\u4e0d\u7b26\u5408\u6761\u4ef6\u7684\u884c\u3002<\/p>\n
      \u53bb\u9664\u5f02\u5e38\u884c\u540e\u5982\u4f55\u9a8c\u8bc1\u6570\u636e\u7684\u5b8c\u6574\u6027\uff1f<\/strong>
      \u5728\u53bb\u9664\u5f02\u5e38\u884c\u4e4b\u540e\uff0c\u786e\u4fdd\u6570\u636e\u7684\u5b8c\u6574\u6027\u81f3\u5173\u91cd\u8981\u3002\u53ef\u4ee5\u4f7f\u7528info()<\/code>\u65b9\u6cd5\u68c0\u67e5\u5269\u4f59\u6570\u636e\u7684\u7ed3\u6784\u548c\u7f3a\u5931\u503c\u60c5\u51b5\u3002\u6b64\u5916\uff0c\u7ed8\u5236\u6570\u636e\u7684\u5206\u5e03\u56fe\uff08\u5982\u76f4\u65b9\u56fe\u6216\u7bb1\u5f62\u56fe\uff09\u6709\u52a9\u4e8e\u76f4\u89c2\u4e86\u89e3\u6570\u636e\u7684\u53d8\u5316\uff0c\u786e\u4fdd\u7ecf\u8fc7\u6e05\u6d17\u540e\u7684\u6570\u636e\u7b26\u5408\u9884\u671f\u7684\u6807\u51c6\u548c\u6a21\u5f0f\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"\u5728 Python \u4e2d\u53bb\u9664\u8868\u683c\u4e2d\u7684\u5f02\u5e38\u884c\u53ef\u4ee5\u901a\u8fc7\u591a\u79cd\u65b9\u6cd5\u6765\u5b9e\u73b0\uff0c\u5e38\u7528\u7684\u65b9\u6cd5\u6709\u4f7f\u7528 pandas \u5e93\u3001\u6570\u636e\u6e05\u6d17\u548c\u8fc7 […]","protected":false},"author":3,"featured_media":1180646,"comment_status":"closed","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[37],"tags":[],"acf":[],"_links":{"self":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/1180621"}],"collection":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/users\/3"}],"replies":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/comments?post=1180621"}],"version-history":[{"count":"1","href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/1180621\/revisions"}],"predecessor-version":[{"id":1180647,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/posts\/1180621\/revisions\/1180647"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media\/1180646"}],"wp:attachment":[{"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/media?parent=1180621"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/categories?post=1180621"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/docs.pingcode.com\/wp-json\/wp\/v2\/tags?post=1180621"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}