过采样与欠采样-解决机器学习之样本不均衡问题(python之sklearn实现).pdfVIP

六六 DATA 过采样与欠采样之sklearn 实现-解决机器学习之样本不均衡 问题 #查看各个标签的样本量--有多不均衡 from collections import Counter Counter(y_old.loss_judge) y_old.shape y_() y_old.describe() ######随机过采样###### # 使用imblearn 进行随机过采样 from imblearn.over_sampling import RandomOverSampler ros = RandomOverSampler(random_state=0) X_resampled, y_resampled = ros.fit_resample(X_old, y_old) #查看结果 Counter(y_resampled.loss_judge) ######过采样代表性算法-SMOTE###### # SMOTE 过采样 from imblearn.over_sampling import SMOTE X_resampled, y_resampled = SMOTE().fit_resamp