house_prices.py 文件源码

def skew_correction(df, numerical_features):
        # Skew correction
        skewed_feats = df[numerical_features].apply(lambda x: skew(x.dropna()))  # compute skewness
        skewed_feats = skewed_feats[skewed_feats > 0.75]
        skewed_feats = skewed_feats.index
        df.loc[:, tuple(skewed_feats)] = np.log1p(np.asarray(df[skewed_feats], dtype=float))
        # df[skewed_feats] = np.log1p(np.asarray(df[skewed_feats], dtype=float))