Credit scoring to classify consumer loan using machine learning

Azaria Natasha*, Dedy Dwi Prastyo, Suhartono

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Citations (Scopus)

Abstract

Credit risk is a potential loss caused by the inability of the debtor to the obligations of debt repayment of either principal or interest debt or both. The classification of credit risk in the financial sector has an essential role in mapping the consumer risk. The wrong classification raises chain effects such as the emergence of bad credit, disruption of financial stability, which lead to banking losses. Classification in credit risk categories the customer loan into two types, good payers or bad payers (default). The aim of this research is to classify consumer's risk to minimize the risk of default. In the past decades, credit scoring using parametric techniques has been applied in the financial field, namely Discriminant Analysis and Binary Logistic Regression. In the last two decades, the non-parametric machine learning approaches, such as Neural Network and Support Vector Machine. Recently, Deep Learning era has been studied widely in credit scoring, like Deep Neural Network. This study is comparing the performance of several methods of non-parametric machine learning and parametric statistics to classify customer loans. Best method to classify customer loan is DNN with number of neuron in h1 = 10, h2 = 3 with value of AUC is 0.638 in testing dataset.

Original languageEnglish
Title of host publication2nd International Conference on Science, Mathematics, Environment, and Education
EditorsNurma Yunita Indriyanti, Murni Ramli, Farida Nurhasanah
PublisherAmerican Institute of Physics Inc.
ISBN (Electronic)9780735419452
DOIs
Publication statusPublished - 18 Dec 2019
Event2nd International Conference on Science, Mathematics, Environment, and Education, ICoSMEE 2019 - Surakarta, Indonesia
Duration: 26 Jul 201928 Jul 2019

Publication series

NameAIP Conference Proceedings
Volume2194
ISSN (Print)0094-243X
ISSN (Electronic)1551-7616

Conference

Conference2nd International Conference on Science, Mathematics, Environment, and Education, ICoSMEE 2019
Country/TerritoryIndonesia
CitySurakarta
Period26/07/1928/07/19

Fingerprint

Dive into the research topics of 'Credit scoring to classify consumer loan using machine learning'. Together they form a unique fingerprint.

Cite this