windowtest: Window length tests

In jeffniv/math5973NivProj2: A Simple Framework for Singular Spectrum Analysis

Description

Plots Lag-correlation, Autocorrelation, and Partial Autocorrelation as a heuristic for determining optimal window length.

Usage

windowtest(ssa)

Arguments

ssa

Details

When choosing a window length, one should consider the lengths associated with the highest and lowest correlation values in the top plot, this tells the user the approx. period of oscillatory data. The autocorrelation plot in the middle gives significance for relationships in the time series, steps in the autocorrelation plot signify signal lengths. In the partial autocorrelation plot, the values in the index behind the values near zero are considered significant length for signal lengths.

Value

None

Examples

obj=ssa(co2,18)
windowtest(obj)

jeffniv/math5973NivProj2 documentation built on April 30, 2020, 8:11 p.m.