Detect significant predictors from bootstrap confidence intervals

This function converts the matrix returned by confints.bootpls into a logical indicator of significance. A predictor is marked as significant when the selected confidence interval lies entirely on one side of the reference value.

Usage

confints2signifind(ic_bootobject, typeIC, threshold = 0)

Arguments

ic_bootobject

a matrix returned by confints.bootpls.

typeIC

type of confidence interval to use. Defaults to "BCa" when BCa limits are available and to "Percentile" otherwise.

threshold

reference value to test against. Defaults to 0.

Value

Named logical vector. TRUE means that the selected confidence interval excludes threshold; FALSE means that it contains threshold; NA means that the interval could not be evaluated.

See also

confints.bootpls and signpred.

Author

Frédéric Bertrand
frederic.bertrand@lecnam.net
https://fbertran.github.io/homepage/

Examples

# \donttest{
data(Cornell)
set.seed(250)
modpls <- plsR(Y ~ ., data = Cornell, 3)
#> ____************************************************____
#> ____Component____ 1 ____
#> ____Component____ 2 ____
#> ____Component____ 3 ____
#> ____Predicting X without NA neither in X nor in Y____
#> ****________________________________________________****
#> 
Cornell.bootYT <- bootpls(modpls, typeboot = "fmodel_np", R = 250, verbose = FALSE)
temp.ci <- confints.bootpls(Cornell.bootYT, indices = 2:8, typeBCa = FALSE)
confints2signifind(temp.ci)
#>    X1    X2    X3    X4    X5    X6    X7 
#>  TRUE  TRUE  TRUE  TRUE FALSE  TRUE  TRUE 
# }