getUpsilons: Compute residual matrix for boosting object
In davidruegamer/iboost: Inference for model-based boosting

Description Usage Arguments Value Examples

Compute residual matrix for boosting object

1 2	getUpsilons(mod, otherSelCourse = NULL, last = FALSE, listFun = kronecker, returnUpsOnly = TRUE)

`mod`	an object of class `mboost`.
`otherSelCourse`	either `NULL` or `numeric`. If `NULL` Upsilon is computed for `selected(mod)`, otherwise for otherSelCourse
`last`	logical; whether to return only Upsilon for the last selection step. If FALSE, a list of Upsilons for all iterations is returned
`listFun`	function; used to combine design matrices, which are build with baselearner operators such as `%X%`.
`returnUpsOnly`	logical; if FALSE, the hat matrices are also returned.

Returns a list of length mstop(mod)+1 with residual matrices of size length(mod$response).

if(require("mboost") & require("TH.data")){
data("bodyfat", package = "TH.data")
mod <- mboost(DEXfat ~ bols(waistcirc) + bbs(hipcirc), data = bodyfat)
Ups <- getUpsilons(mod)
# check result
cbind(mod[100]$resid()[1:10],
(Ups[[length(Ups)]]%*%(mod$response-mod$offset))[1:10]
)
}