MakeEmbedded: Embedding of multivariate time series

In gbonte/gbcode: Code from the handbook "Statistical foundations of machine learning"

Embedding of multivariate time series

Description

Embed a multivariate time series in input output form

Usage

MakeEmbedded(ts, n, delay, hor = 1, w = 1)

Arguments

n	[no.var]: vector of embedding orders
delay	[no.var]: vector of delays
hor	[length(w)]: vector of predicted horizons (hor=1 boils down to one-step-ahed prediction)
ts:	multivariate time series [no. observations,no. variates]
w:	index of variables appearing in $out

Value

list with

• inp: embedded inputs of all variables (n[i] columns for series ts[i])

• out: outputs of variates whose index is in w

Author(s)

Gianluca Bontempi [e-mail]

References

mlg.ulb.ac.be

Examples

TS<-array(1:25,c(5,5))
## embeds a 5 variate time series with embedding orders equal to 2 for all variables for one-step ahead prediction
MakeEmbedded(TS,n=rep(2,5),delay=rep(0,5),hor=rep(1,5),w=1:5)

gbonte/gbcode documentation built on Aug. 30, 2024, 1:11 a.m.