Home › Stats-TimeSeries › Seasonality

SeasonalDecomposition

Classical decomposition of a series into trend, seasonal pattern and residual, by moving averages.

public static class SeasonalDecomposition

Example — three years of quarterly figures, a rising level with a strong second quarter.

using Lodestar.Stats.TimeSeries;

double[] quarterly = [10.0, 14.0, 8.0, 12.0, 11.0, 15.0, 9.0, 13.0, 12.0, 16.0, 10.0, 14.0];

SeasonalComponents parts = SeasonalDecomposition.Decompose(quarterly, period: 4);

double secondQuarter = Math.Round(parts.Seasonal[1], 4);  // => 3.125

Remarks — classical decomposition, not STL: the trend is a centred moving average and the seasonal pattern the average detrended value at each phase. STL's loess smoothing is a different algorithm and is not in this package.

Applies to — net10.0, netstandard2.0.

See also — the seasonality index, Stationarity.

Members

Member	What it does
SeasonalDecomposition.Decompose	Splits a series into its trend, seasonal and residual components.