Using data mining approaches to build credit scoring model: Case study — Implementation of credit scoring model in microfinance institution

摘要

The aim of this paper is to present a Credit Scoring Model applied by one Microfinance Institution in Bosnia and Herzegovina as well as to show how the most relevant attributes for its implementation were identified. The purpose of created Credit Scoring Model was to predict default clients and reduce credit risk of Microfinance Institution by applying data mining algorithm in order to find patterns for recognition of default clients and, thus, support decision making process of credit approval. Credit Scoring Model was build using Oracle Data Miner software package that uses Generalized Linear Model for classification. Created model showed great predictive confidence and accuracy, but also gave trustworthy results regarding feature selection, so the Microfinance institution decided to adopt this model as help in decision making process.