balanceMultiProbs: Simple approach to balance the probabilities from an...

In wtcooper/modUtils: Collection of utils for model building and tuning.

Description

Simple approach to balance the probabilities from an inbalanced multiclass/multinomial model. Divides the probability by the class frequency used in the training dataset. Note: this calculates based on the named columns and the named frequencies matching so makes sure they match ok

Usage

balanceMultiProbs(probs, freqs)

Arguments

probs	a dataframe of the probabilities, each column with the class name in factor order
freqs	named proportions from the training data as from prop.table(table(target))

wtcooper/modUtils documentation built on May 4, 2019, 11:59 a.m.