Open AccessData Mining Application in Predicting Bank Loan Defaulters
Author(s) -
Jeremy Nixon,
Ashenafi Wubshet Desta
Publication year - 2020
Publication title -
international journal of innovative technology and exploring engineering
Language(s) - English
Resource type - Journals
ISSN - 2278-3075
DOI - 10.35940/ijitee.d2037.029420
Subject(s) - c4.5 algorithm , decision tree , loan , computer science , data mining , naive bayes classifier , allotment , key (lock) , sample (material) , process (computing) , machine learning , finance , support vector machine , business , ecology , chemistry , computer security , chromatography , biology , operating system
Data mining is the key tools for discoveries of knowledge from large data set. Nowadays, most of the organizations using this technology to maintain their data. This paper focuses on the Bank sector in Risk management specifically, detecting Bank loan defaulters through the data mining application to examine the patterns of different attribute which would contribute for detecting and predicting defaulters thus preventing wrong loans. This process can be done without change the current systems and the data. Then it helps to distinguish borrowers who repay loans promptly from those who don’t and avoid wrong loan allotment. In order to show the results of the study Classification model is implemented in order to find interesting patterns among attributes of customer. A total of 20461 sample data were taken by data base admin randomly from 3 consecutive years from the Bank database to build and test the model. In this research we used Classification model of decision tree and Naïve Bayes in Weka 3.7 tool for experiments. Modeling methodology applied to this paper was CIRSP-DM (Cross Industry Standard for Data Mining), which involves business understanding, data understanding, data preparation, model building, evaluation and deployment. Decision tree classifications with J48 implementation with 8 experiments were performed. Two experiments with different parameters were made for Naïve Bayes. Finally, evaluation and analysis of the models were performed then given a best solution to predict the defaulters.