评分卡模型开发-定性指标筛选

library(InformationValue)
library(klaR)
credit_risk<-ifelse(train_kfolddata[,"credit_risk"]=="good",0,1)
#将违约状态变量用0和1表示，1表示违约。
tmp<-train_kfolddata[,-21]
data<-cbind(tmp,credit_risk)
data<-as.data.frame(data)

factor_vars<-c("status","credit_history","purpose","savings","employment_duration",
               "personal_status_sex","other_debtors","property",
               "other_installment_plans","housing","job","telephone","foreign_worker")
#获取所有名义变量
all_iv<-data.frame(VARS=factor_vars,IV=numeric(length(factor_vars)),
                   STRENGTH=character(length(factor_vars)),stringsAsFactors = F)
#初始化待输出的数据框
for(factor_var in factor_vars)
{
  all_iv[all_iv$VARS==factor_var,"IV"]<-InformationValue::IV(X=
                                                               data[,factor_var],Y=data$credit_risk)  
  #计算每个指标的IV值
  all_iv[all_iv$VARS==factor_var,"STRENGTH"]<-attr(InformationValue::IV(X=
                                                                          data[,factor_var],Y=data$credit_risk),"howgood")  
  #提取每个IV指标的描述
}
all_iv<-all_iv[order(-all_iv$IV),]    #排序IV