dataframe字符串型空白值处理；随机、分层、过采样

pandas字符串型空白值处理

import numpy as np
all_datass.replace(to_replace&＃61;r&＃39;^\s*$&＃39;,value&＃61;np.nan,regex&＃61;True,inplace&＃61;True)
all_datass &＃61;all_datass.dropna()


pandas随机、分层、过采样

随机&＃xff1a;pandas sample

DataFrame.sample(n&＃61;None, frac&＃61;None, replace&＃61;False, weights&＃61;None, random_state&＃61;None, axis&＃61;None)[source]


分层&＃xff1a;可以是几个特定的类别再随机sample后组合

df[df["s"]&＃61;&＃61;"a"].sample()
df[df["s"]&＃61;&＃61;"b"].sample()
df[df["s"]&＃61;&＃61;"c"].sample()


过采样:&＃xff08;上&＃xff0c;下采样&＃xff09;

from imblearn.over_sampling import SMOTE
X_resampled_smote, y_resampled_smote &＃61; SMOTE(sampling_strategy&＃61;0.05).fit_sample(X, y)
sorted(Counter(y_resampled_smote).items())