Imputer

class sktime.transformations.series.impute.Imputer(method, random_state=None, value=None, forecaster=None, missing_values=None)[source]

Missing value imputation

Parameters
  • method (Method to fill values.) –

    • “drift” : drift/trend values by sktime.PolynomialTrendForecaster()

    • ”linear” : linear interpolation, by pd.Series.interpolate()

    • ”nearest” : use nearest value, by pd.Series.interpolate()

    • ”constant” : same constant value (given in arg value) for all NaN

    • ”mean” : pd.Series.mean()

    • ”median” : pd.Series.median()

    • ”backfill”/”bfill” : adapted from pd.Series.fillna()

    • ”pad”/”ffill” : adapted from pd.Series.fillna()

    • ”random” : random values between pd.Series.min() and .max()

    • ”forecaster” : use an sktime Forecaster, given in arg forecaster

  • missing_values (int/float/str, optional) – The placeholder for the missing values. All occurrences of missing_values will be imputed. Default, None (np.nan)

  • value (int/float, optional) – Value to fill NaN, by default None

  • forecaster (Any Forecaster based on sktime.BaseForecaster, optinal) – Use a given Forecaster to impute by insample predictions. Before fitting, missing data is imputed with method=”ffill”/”bfill” as heuristic.

  • random_state (int/float/str, optional) – Value to set random.seed() if method=”random”, default None

__init__(method, random_state=None, value=None, forecaster=None, missing_values=None)[source]

Initialize self. See help(type(self)) for accurate signature.