Stats TimeSeries seasonality - CyrilB1531/lodestar GitHub Wiki

HomeStats-TimeSeries

Seasonality — Lodestar.Stats.TimeSeries

Classical decomposition of a series into a trend, a seasonal pattern and a residual, by moving averages, at statsmodels 0.15.0's seasonal_decompose parity.

type carries
SeasonalDecomposition SeasonalDecomposition.Decompose, the one entry point
SeasonalDecompositionOptions the model, the filter's sides and the trend extrapolation
SeasonalComponents the three components, each the length of the series
SeasonalModel additive or multiplicative

The period is required: the reference infers it from a pandas index, which a span does not have. STL, the loess-based decomposition, is a different algorithm and is not here.

The Python equivalence table maps each statsmodels call to its counterpart here.