pull out attribute usage stats from c5.0 objects in R

c50_usage_df.R

c5_usage_df <- function(x) {
  library(stringr)
  xt <- substr(x$output, 
               str_locate(x$output, "usage:")[2] + 4, 
               str_locate(x$output, 'Time')[1] - 1)
  
  xt <- gsub('\\n', '', xt)
  xt <- gsub('\\t', ', ', xt)
  xu <- as.data.frame(matrix(unlist(strsplit(xt, ", ")),
                             ncol =2, byrow=T),
                      stringsAsFactors = FALSE)
  names(xu) <- c('use_percentage', 'covariate_name')
  xu$use_percentage <- gsub('%', "", xu$use_percentage)
  xu$use_percentage <- gsub(' ', "", xu$use_percentage)
  xu$use_percentage <- as.numeric(xu$use_percentage) 
  return(xu)
}