EWSmethods (version 1.3.1)

tuneII: Information Imbalance Across Alphas

Description

Estimates the information imbalance of two hypothesised linked system measurements using distance ranks.

Usage

tuneII(columns, target, tau, alphas, k = 1, method = "euclidean")

Value

A dataframe of alphas and the estimate information imbalance

Arguments

columns

Numeric matrix of hypothesised driving variable measurements. If univariate, call `embedTS(X)` prior to calling `II()`.

target

Numeric matrix of hypothesised response variable measurements. If univariate, call `embedTS(Y)` prior to calling `II()`.

tau

Numeric. Time lag of information transfer between X and Y.

alphas

Numeric vector. Range of X scaling parameters bewtween `0` & `1` inclusive. If information imbalance is minimised at an `alpha` > 0, this may be indicative of Granger causality.

k

Numeric. Number of nearest neighbours when estimating ranks.

method

String. Distance measure to be used - defaults to `euclidean` but see `?dist` for options.

Examples

#Load the multivariate simulated
#dataset `simTransComms`

data(simTransComms)

#Embed the spp_2 and spp_5 of the third community

embedX <- embed_ts(X = simTransComms$community3[,c("time","spp_2")],
E = 5, tau = 1)

embedY <- embed_ts(X = simTransComms$community3[,c("time","spp_5")],
E = 5, tau = 1)

alphas <- seq(from = 0, to = 1, by = 0.1)

# \donttest{
#if parallelisation desired,
#this can be achieved using the
#below code
cl <- parallel::makeCluster(2)
# }

# \donttest{
doParallel::registerDoParallel(cl)
# }

#Estimate the forward information imbalance
#between spp_2 and spp_5

egII_for <- tuneII(target = embedX[,-1], columns = embedY[,-1],
tau = 1, alphas = alphas, k = 5)

#Estimate the reverse information imbalance
#between spp_2 and spp_5

egII_rev <- tuneII(target = embedY[,-1], columns = embedX[,-1],
tau = 1, alphas = alphas, k = 5)

# \donttest{
parallel::stopCluster(cl)
# }