Credit Risk Scorecard

The `scorecard` package makes the development of credit risk scorecard easier and efficient by providing functions for some common tasks, such as data partition, variable selection, woe binning, scorecard scaling, performance evaluation and report generation. These functions can also used in the development of machine learning models. The references including: 1. Refaat, M. (2011, ISBN: 9781447511199). Credit Risk Scorecard: Development and Implementation Using SAS. 2. Siddiqi, N. (2006, ISBN: 9780471754510). Credit risk scorecards. Developing and Implementing Intelligent Credit Scoring.

The goal of scorecard package is to make the development of traditional credit risk scorecard model easier and efficient by providing functions for some common tasks.

• data partition (split_df)
• variable selection (iv, var_filter)
• weight of evidence (woe) binning (woebin, woebin_plot, woebin_adj, woebin_ply)
• scorecard scaling (scorecard, scorecard_ply)
• performance evaluation (perf_eva, perf_psi)

Installation

• Install the release version of scorecard from CRAN with:

install.packages("scorecard")

• Install the latest version of scorecard from github with:

# install.packages("devtools")
devtools::install_github("shichenxie/scorecard")

Example

This is a basic example which shows you how to develop a common credit risk scorecard:

# Traditional Credit Scoring Using Logistic Regression
library(scorecard)
 
# data prepare ------
# load germancredit data
data("germancredit")
 
# filter variable via missing rate, iv, identical value rate
dt_s = var_filter(germancredit, y="creditability")
 
# breaking dt into train and test
dt_list = split_df(dt_s, y="creditability", ratio = 0.6, seed = 30)
train = dt_list$train; test = dt_list$test;
 
# woe binning ------
bins = woebin(dt_s, y="creditability")
# woebin_plot(bins)
 
# binning adjustment
# # adjust breaks interactively
# breaks_adj = woebin_adj(dt_s, "creditability", bins) 
# # or specify breaks manually
breaks_adj = list(
  age.in.years=c(26, 35, 40),
  other.debtors.or.guarantors=c("none", "co-applicant%,%guarantor"))
bins_adj = woebin(dt_s, y="creditability", breaks_list=breaks_adj)
 
# converting train and test into woe values
train_woe = woebin_ply(train, bins_adj)
test_woe = woebin_ply(test, bins_adj)
 
# glm ------
m1 = glm( creditability ~ ., family = binomial(), data = train_woe)
# summary(m1)
 
# # Adjusting for oversampling (support.sas.com/kb/22/601.html)
# library(data.table)
# p1=0.03; r1=0.3
# dt_woe = dt_woe[, weight := ifelse(y==1, p1/r1, (1-p1)/(1-r1) )]
# 
# fmla = as.formula(paste("y ~", paste(names(dt_woe)[2:21], collapse="+")))
# m1 = glm(fmla, family = binomial(), data = dt_woe, weights = weight)
 
 
# Select a formula-based model by AIC (or by LASSO)
m_step = step(m1, direction="both", trace = FALSE)
m2 = eval(m_step$call)
# summary(m2)
 
# performance ks & roc ------
# predicted proability
train_pred = predict(m2, train_woe, type='response')
test_pred = predict(m2, test_woe, type='response')
# performance
train_perf = perf_eva(train$creditability, train_pred, title = "train")
test_perf = perf_eva(test$creditability, test_pred, title = "test")
 
# score ------
card = scorecard(bins_adj, m2)
# credit score
train_score = scorecard_ply(train, card, print_step=0)
test_score = scorecard_ply(test, card, print_step=0)
 
# psi
perf_psi(
  score = list(train = train_score, test = test_score),
  label = list(train = train$creditability, test = test$creditability)
)