biogram (version 1.1)

calc_ig_single: Calculate IG for single feature

Description

Computes information gain of single feature and target vector.

Usage

calc_ig_single(feature, target_b, len_target, pos_target, ES)

Arguments

feature

feature vector.

target_b

target in bits (as per as.bit).

len_target

length of target vector.

pos_target

number of positive cases in target vector.

ES

numeric value of target entropy.

Value

• a single numeric value - information gain in nats.

Details

Input looks strange, but the function was designed to be as fast as possible subroutine of calc_ig and generally should not be directly called by user.

Examples

tar <- sample(0L:1, 100, replace = TRUE)
feat <- sample(0L:1, 100, replace = TRUE)
prop <- c(100 - sum(tar), sum(tar))/100
entr <- - sum(prop*log(prop))
library(bit) #used to code vector as bit
calc_ig_single(feat, as.bit(tar), 100, sum(tar), entr)