qolmat.imputations.imputers.ImputerResiduals

class qolmat.imputations.imputers.ImputerResiduals(period: int = 1, groups: Tuple[str, ...] = (), model_tsa: Optional[str] = 'additive', extrapolate_trend: Optional[Union[int, str]] = 'freq', method_interpolation: Optional[str] = 'linear')

Residual imputer.

This class implements an imputation method based on a STL decomposition. The series are de-seasonalised, de-trended, residuals are imputed, then residuals are re-seasonalised and re-trended.

Parameters

groups: Tuple[str, …]

List of column names to group by, by default []

periodint

Period of the series. Must be used if x is not a pandas object or if the index of x does not have a frequency. Overrides default periodicity of x if x is a pandas object with a timeseries index.

model_tsaOptional[str]

Type of seasonal component “additive” or “multiplicative”. Abbreviations are accepted. By default, the value is set to “additive”

extrapolate_trendint or ‘freq’, optional

If set to > 0, the trend resulting from the convolution is linear least-squares extrapolated on both ends (or the single one if two_sided is False) considering this many (+1) closest points. If set to ‘freq’, use freq closest points. Setting this parameter results in no NaN values in trend or resid components.

method_interpolationstr

method for the residuals interpolation

Examples

>>> import numpy as np
>>> import pandas as pd
>>> from qolmat.imputations.imputers import ImputerResiduals
>>> np.random.seed(100)
>>> df = pd.DataFrame(index=pd.date_range("2015-01-01", "2020-01-01"))
>>> mean = 5
>>> offset = 10
>>> df["y"] = np.cos(df.index.dayofyear / 365 * 2 * np.pi - np.pi) * mean + offset
>>> trend = 5
>>> df["y"] = df["y"] + trend * np.arange(0, df.shape[0]) / df.shape[0]
>>> noise_mean = 0
>>> noise_var = 2
>>> df["y"] = df["y"] + np.random.normal(noise_mean, noise_var, df.shape[0])
>>> mask = np.random.choice([True, False], size=df.shape)
>>> df = df.mask(mask)
>>> imputor = ImputerResiduals(period=365, model_tsa="additive")
>>> imputor.fit_transform(df)
                    y
2015-01-01   1.501210
2015-01-02   5.691061
2015-01-03   4.404106
2015-01-04   3.531540
2015-01-05   3.129532
...               ...
2019-12-28  10.288054
2019-12-29  10.632659
2019-12-30  14.900671
2019-12-31  12.957837
2020-01-01  12.780517

[1827 rows x 1 columns]

__init__(period: int = 1, groups: Tuple[str, ...] = (), model_tsa: Optional[str] = 'additive', extrapolate_trend: Optional[Union[int, str]] = 'freq', method_interpolation: Optional[str] = 'linear')

Examples using qolmat.imputations.imputers.ImputerResiduals

Benchmark for time series

Benchmark for time series