gfmR: Implements Group Fused Multinomial Regression

Software to implement methodology to preform automatic response category combinations in multinomial logistic regression. There are functions for both cross validation and AIC for model selection. The method provides regression coefficient estimates that may be useful for better understanding the true probability distribution of multinomial logistic regression when category probabilities are similar. These methods are not recommended for a large number of predictor variables.

install.packages("gfmR")

Try the gfmR package in your browser