A Deep Neural Network (DNN) based classification model in application to loan default prediction

Selçuk Bayraci and Orkun Susuz
Additional contact information
Selçuk Bayraci: R&D Center, C/S Information Technologies, Istanbul, Turkey
Orkun Susuz: R&D Center, C/S Information Technologies, Istanbul, Turkey

Theoretical and Applied Economics, 2019, vol. XXVI, issue 4(621), Winter, 75-84

Abstract: In this study, we applied a Deep Neural Networks (DNN) based classification model along with the conventional classification methods (Logistic Regression, Decision Tree, Naïve Bayes and Support Vector Machines) on a two distinct datasets containing characteristics of the loan clients in a medium-sized Turkish commercial bank. Python programming language and libraries (Sklearn, Tensorflow and Keras) have been used in data cleaning, data preparation, feature engineering and model implementation processes. Our empirical findings document that the accuracy of the deep learning classification model increases with the size of the dataset, implying that the deep learning models might yield better results than regression-based models in more complex datasets.

Keywords: data analytics; credit scoring; deep learning; risk management. (search for similar items in EconPapers)
Date: 2019
References: View references in EconPapers View complete reference list from CitEc
Citations: View citations in EconPapers (3)

Downloads: (external link)
http://store.ectap.ro/articole/1421.pdf (application/pdf)
http://www.ectap.ro/articol.php?id=1421&rid=137 (text/html)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:agr:journl:v:xxvi:y:2019:i:4(621):p:75-84

Access Statistics for this article

Theoretical and Applied Economics is currently edited by Mircea Dinu

More articles in Theoretical and Applied Economics from Asociatia Generala a Economistilor din Romania / Editura Economica Contact information at EDIRC.
Bibliographic data for series maintained by Mircea Dinu ().