Kaggle案例精选——电信客户流失预测（Telecom Customer Churn Prediction）Part Four：模型表现对比

最新推荐文章于 2024-07-01 19:35:08 发布

「已注销」

最新推荐文章于 2024-07-01 19:35:08 发布

阅读量2.6k

点赞数 1

分类专栏： Kaggle数据分析项目文章标签：数据分析数学建模 python kaggle

本文链接：https://blog.csdn.net/Haoyu_xie/article/details/108575919

版权

本文探讨了Kaggle上的电信客户流失预测案例，重点分析了不同模型的表现。通过构建指标计算函数，展示了模型的评价指标，包括混淆矩阵、ROC曲线和精确度召回曲线，以全面评估模型的性能。

摘要由CSDN通过智能技术生成

6 Model Performance：模型表现

构建指标计算函数

from sklearn.metrics import f1_score, cohen_kappa_score, precision_recall_curve, average_precision_score

# 设置模型报告表格
def model_report_df(model, training_x, testing_x, training_y, testing_y, name):
    model.fit(training_x, training_y)
    preds = model.predict(testing_x)
    accracy = accuracy_score(testing_y, preds)
    recallscore = recall_score(testing_y, preds)
    precision = precision_score(testing_y, preds)
    roc_auc = roc_auc_score(testing_y, preds)
    f1score = f1_score(testing_y, preds)
    kappa_metric = cohen_kappa_score(testing_y, preds)

    df = pd.DataFrame({
   
        'Model': [name],
        'Accuracy':[accracy],
        'Recall_score':[recallscore],
        'Precision':[precision],
        'F1_score':[f1score],
        'Area_under_curve':[roc_auc],
        'Kappa_metric':[kappa_metric]
    })
    return df

6.1 模型表现指标计算

# 每个模型的输出值计算
model1 = model_report_df(logit, train_X, test_X, train_Y, test_Y,
                         'Logistic Regression(Baseline_model)')
model2 = model_report_df(logit_smote, os_smote_X, test_X, os_smote_Y, test_Y,
                         'Logistic Regression(SMOTE)')
model3 = model_report_df(logit_rfe, train_X, test_X, train_Y, test_Y,
                         'Logistic Regression(RFE)')

DTree = DecisionTreeClassifier(max_depth=9, random_state=123, splitter='best', criterion='gini')
model4 = model_report_df(DTree, train_X, test_X, train_Y, test_Y,
                         'Decision Tree')
model5 = model_report_df(knn, os_smote_X, test_X, os_smote_Y, test_Y,
                         'KNN Classifier')

rfc = RandomForestClassifier(n_estimators=1000, random_state=123, max_depth=9, criterion='gini')
model6 = model_report_df(rfc, train_X, test_X, train_Y, test_Y