EmbedFn: Find optimal simplex 'embedding' dimension

In nwfscNLTS: Non-linear time series

Description

Function that returns the embedding dimention for simplex that maximizes cross-validation correlation

Usage

  EmbedFn(Y, PredInterval, Candidates = 1:5)

Arguments

Y	Time series
PredInterval	Prediction interval, i.e. 1 represents predicting one observation ahead
Candidates	Candidate embedding dimensions

Value

Returns an integer from the 'Candidates' which maximizes crossvalidation correlation

Author(s)

James Thorson

References

Sugihara, G., and May, R.M. 1990. Nonlinear forecasting as a way of distinguishing chaos from measurement error in time series. Nature 334: 734-741.

Examples

# Time series length
Nobs = 100

# Generate Lotka-Volterra timeseries
Y = SimPredPreyFn(Nobs=Nobs, Nt=100, A=0.4, B=0.5, C=0.2, E=1)$Y

# Simplex
Output = NltsFn(Y, PredInterval=2, Nembed=1, Method="Simplex")
EmbedFn(c(Y,NA,NA), PredInterval=2, Candidates=1:10)

nwfscNLTS documentation built on May 2, 2019, 4:31 p.m.