beset: beset: Best Subset Predictive Modeling

In jashu/beset: Best Subset Predictive Modeling

beset: Best Subset Predictive Modeling

Description

beset is a portmanteau of BEst subSET, which references the overall objective of this package: to identify the best subset of variables for a predictive model. To learn more about beset, start with the vignettes: browseVignettes(package = "beset")

Overarching goals

1. Provide a fast and easy way to cross-validate GLMs.

2. Establish a common, user-friendly interface for best subset selection that works with several model fitting functions (lm, glm, and glm.nb).

3. Make elastic-net regression more accessible and interpretable by providing a wrapper to glmnet that maintains the same user interface as lm and glm and provides informative plot and summary methods.

Overview of principal functions

beset_glm

Performs best subset selection using repeated cross-validation to find the optimal number of predictors for several families of generalized linear models.

beset_elnet

Enhances elastic-net regression with glmnet by 1) allowing the user to specify a model using R's formula syntax, 2) allowing the user to simultaneously tune both alpha and lambda using cross-validation, and 3) providing a summary output that ranks the relative importance of the predictors that survived shrinkage and gives some statistics indicating how well the model fits the training data and predicts new data.

validate

An S3 object system that makes it easy to obtain cross-validated prediction stats for a previously fit model.

jashu/beset documentation built on April 20, 2023, 5:28 a.m.