Loan Default Predictive Analytics

Ebenezer Owusu, Richard Quainoo, Justice Kwame Appati, Solomon Mensah

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

Abstract

Making accurate loan default projections is difficult for lending organizations. Large quantities of money are owed as loans, resulting in a significant loss to lending companies. The study delves into loan default in online peer-to-peer lending by introducing a model to classify a loan instance as default and fully paid. Due to the imbalance nature of the loan default dataset extracted from Kaggle, an Adaptive Synthetic Sampling (ADASYN) approach is used to balance the data by oversampling the minority class. A Deep Neural Network (DNN) is used for the training and validation needs. The prediction accuracy of 94.1% is obtained. This performance is the highest score across multiple experiments with different batch sizes and epochs. The findings clearly demonstrate that the proposed technique is very promising.

Original languageEnglish
Title of host publicationProceedings - 2022 IEEE World Conference on Applied Intelligence and Computing, AIC 2022
EditorsGeetam S. Tomar, Jagdish Bansal
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages617-622
Number of pages6
ISBN (Electronic)9781509050017
DOIs
Publication statusPublished - 2022
Event2022 IEEE World Conference on Applied Intelligence and Computing, AIC 2022 - Sonbhadra
Duration: 17 Jun 202219 Jun 2022

Publication series

NameProceedings - 2022 IEEE World Conference on Applied Intelligence and Computing, AIC 2022

Conference

Conference2022 IEEE World Conference on Applied Intelligence and Computing, AIC 2022
Country/TerritoryIndia
CitySonbhadra
Period17/06/2219/06/22

Keywords

  • Adaptive synthetic sampling (ADASYN)
  • Deep neural network (DNN)
  • Imbalanced dataset
  • Loan-default

Fingerprint

Dive into the research topics of 'Loan Default Predictive Analytics'. Together they form a unique fingerprint.

Cite this