tslumen.profile.components module¶

Functions for decomposing the time series in trend/seasonality/residual.

tslumen.profile.components.seasonal_split(data: pandas.core.series.Series) → pandas.core.frame.DataFrame[source]¶

Splits the data by season:

Quarterly - Years by Quarters
Monthly - Years by Months
Weekly - Years by Weeks
Daily - Year+Months by Days
Business daily - Week by Day of the Week
Hourly - Year+Month+Day by Hours

Parameters: data (pd.Series) – Time series split.
Returns: Seasonally split data.
Return type: pd.DataFrame

tslumen.profile.components.stl(data: pandas.core.series.Series, period: Optional[int] = None, seasonal: int = 7, trend: Optional[int] = None, low_pass: Optional[int] = None, seasonal_deg: Optional[int] = 0, trend_deg: Optional[int] = 0, low_pass_deg: Optional[int] = 0, robust: bool = False, seasonal_jump: int = 1, trend_jump: int = 1, low_pass_jump: int = 1) → pandas.core.frame.DataFrame[source]¶

Season-Trend decomposition using LOESS.

Returns: DataFrame with 3 columns: trend, seasonality and residual.
Return type: pd.DataFrame