Home › Stats-TimeSeries › Stationarity tests

KpssResult

A KPSS test: the statistic, its tabulated p-value and whether that p-value was clamped.

public sealed class KpssResult

Properties — Statistic is the residual partial-sum statistic over the long-run variance. PValue is the p-value against the null of stationarity, interpolated in Kwiatkowski et al.'s table. LagCount is the window the long-run variance used. CriticalValues holds the critical values at 10 %, 5 %, 2.5 % and 1 %. PValueBound says whether PValue is the table's end rather than an interpolation, and which way the truth lies — a PValueBound.

Example — a statistic below the 10 % critical value comes back as the table's largest p-value, with the direction beside it.

using Lodestar.Stats.TimeSeries;

double[] noise = [0.3, -0.5, 0.9, -0.2, 0.1, -0.8, 0.6, -0.4, 0.2, -0.1, 0.7, -0.6];

KpssResult result = Stationarity.Kpss(noise);

double statistic = Math.Round(result.Statistic, 4);  // => 0.3277
double p = result.PValue;                            // => 0.1
PValueBound bound = result.PValueBound;              // => ActualIsGreater
double tenPercent = result.CriticalValues[0];        // => 0.347

Remarks — there is no public constructor. A result is what Stationarity.Kpss returns. The reference returns the same end value and raises an InterpolationWarning; a library has no warning channel a caller reads, so the direction is a property.

A class rather than a record, for the reason AutocorrelationResult gives.

Applies to — net10.0, netstandard2.0.

See also — Stationarity.Kpss, KpssOptions, the Python equivalence table.