Scorecard_bootstrap: Run bootstrap model fits to assess coefficient distributions.

In Zelazny7/binnr: Scorecard modeling in R

Description

Run bootstrap model fits to assess coefficient distributions.

Arguments

times	number of bootstrap samples to run
bag.fraction	fraction of observations to sample for each bootstrap run
replace	whether to sample with or without replacement
nfolds	number of CV folds to run within each bootstrap fit
upper.limits	maximum coffecient value for each fit
lower.limits	minimum coffecient value for each fit
alpha	mixing paramater between LASSO and Ridge regression. Default 1.
...	other parameters passed on to cv.glmnet

Details

True boostratp samples should be run with bag.fraction=1 and replace=TRUE.

Value

a list with two elements: pvals and coefficients. The former is a vector indicating what proportion of bootstrap model fits each coefficient returned a zero. The latter an nRuns x nVars matrix containing the coefficients of each run.

Zelazny7/binnr documentation built on May 10, 2019, 1:55 a.m.