imputeTransdapter: Impute transdapter

In Boruta: Wrapper Algorithm for All Relevant Feature Selection

Impute transdapter

Description

Wraps the importance adapter to accept NAs in input.

Usage

imputeTransdapter(adapter = getImpRfZ)

Arguments

adapter

importance adapter to transform.

Value

transformed importance adapter which can be fed into getImp argument of the Boruta function.

Note

An all-NA feature will be converted to all zeroes, which should be ok as a totally non-informative value with most methods, but it is not universally correct. Ideally, one should avoid having such features in input altogether.

Examples

## Not run: 
set.seed(777)
data(srx)
srx_na<-srx
# Randomly punch 25 holes in the SRX data
holes<-25
holes<-cbind(
 sample(nrow(srx),holes,replace=TRUE),
 sample(ncol(srx),holes,replace=TRUE)
)
srx_na[holes]<-NA
# Use impute transdapter to mitigate them with internal imputation
Boruta(Y~.,data=srx_na,getImp=imputeTransdapter(getImpRfZ))

## End(Not run)

Boruta documentation built on Nov. 12, 2022, 9:06 a.m.