株価データの非定常性に対するロバスト性

elif SCALING == "RemoveMovingAverageAndStandardLobast":
    df = df - df.rolling(window=PCT_DIFF).mean()
    
    split_date_first = "2019-01-01"
    split_date_second = "2022-01-01"
    train_df__ = df[df.index<split_date_first]
    test_df__ = df[(df.index>=split_date_first) & (df.index<split_date_second)]
    scaler = StandardScaler() 
    train_df__ = pd.DataFrame(scaler.fit_transform(train_df__), 
                            index=train_df__.index, columns=train_df__.columns) # normalize for clustering
    test_df__ = pd.DataFrame(scaler.fit_transform(test_df__), 
                            index=test_df__.index, columns=test_df__.columns) # normalize for clustering
    train_df__ = pd.concat([train_df__,test_df__])

    # split_date = "2022-01-01"
    train_df = df[df.index<split_date_second]
    test_df = df[df.index>=split_date_second]
    scaler = StandardScaler() 
    scaler.fit(train_df)
    test_df = pd.DataFrame(scaler.transform(test_df), 
                            index=test_df.index, columns=test_df.columns) # normalize for clustering
    df = pd.concat([train_df__,test_df])


elif SCALING == "RemoveMovingAverageAndStandard":
    df = df - df.rolling(window=PCT_DIFF).mean()
    
    split_date = "2019-01-01"
    train_df = df[df.index<split_date]
    test_df = df[df.index>=split_date]  
    scaler = StandardScaler() 
    train_df = pd.DataFrame(scaler.fit_transform(train_df), 
                            index=train_df.index, columns=train_df.columns) # normalize for clustering
    test_df = pd.DataFrame(scaler.transform(test_df), 
                            index=test_df.index, columns=test_df.columns) # normalize for clustering
    df = pd.concat([train_df,test_df])