orderpredictors: Order predictors based on entropy

In romain-louvet/DotRules: Dictionary of trusted rules for LUCC simulation

Description

Sort columns by increasing entropy values based on the factors of the response variable.

Usage

orderpredictors(data, x, y)

Arguments

data	Dataframe
x	vector of variable names of predictors
y	response variable

Details

The mean entropy value is calculated for each variables based on every outcome in the response variable. Then this value is used for sorting the predictors.

Value

A vector of predictor names ordered by increasing entropy and corresponding entropy values

Examples

data('lulc_data')

x <- c('maxlike_1985','DFU_85','maxlike_1991','DFU_91','maxlike_2000','DFU_00',
      'F00_3by3_W','F00_3by3_R','F00_3by3_B','F00_3by3_A','F00_3by3_U')
y <- 'maxlike_2015'

orderedx <- orderpredictors(lulc_data, x, y)

lulc_model_orderedx <- dotrules(lulc_data,orderedx$x,y, sampsize =.1)

romain-louvet/DotRules documentation built on May 27, 2019, 1:47 p.m.