Home  Journals  Financial Analytics: Science and Experience  6(336) - 2017 June

Adaptation of models predicting customers' creditworthiness in line with new incoming data on clients

Vol. 10, Iss. 6, JUNE 2017

Article PDF Version

Received: 20 December 2016

Received in revised form: 17 March 2017

Accepted: 24 March 2017

Available online: 15 June 2017

Subject Heading: MONITORING AND PREDICTION OF BANKING RISKS

JEL Classification: C02, C53, G21

Pages: 663-674

https://doi.org/10.24891/fa.10.6.663

Importance Machine learning methods are used to predict repayment of loans by borrowers. It is an acute task to update the structure of the aggregate forecast after some time in order to adapt to new clients' characteristics as well as to provide sufficient prediction accuracy.
Objectives The aim is to provide the structure updating of the aggregate forecasting method to improve the prediction accuracy.
Methods In this paper, we used the machine learning methods with different classifiers based on a neural network, logistic regression, discriminant analysis, naive Bayes classifier, Support Vector Machines, etc. To adjust parameters of the models, pseudo-gradient procedure is used. The quality of the model structure obtained with updated parameters is assessed by the average square error in the control sample.
Results We developed a model structure updating method for forecasting customers' loan repayment, the effectiveness of which has been confirmed by practical tests based on the data according to Russian borrowers.
Conclusions and Relevance Using the pseudo-gradient procedure of adjusting the parameters of the chosen model provides an accurate prediction for a certain period of time. However, after significant changes in the credit situation, this model is not able to provide a sufficient accuracy of forecast for any values of its parameters. Therefore, from time to time it is needed to change the structure of the model to provide sufficient prediction accuracy to get more justified decisions about granting credits to new customers on the new terms. The proposed method can be applied in automated systems to support decision-making about granting loans in the banking sector.

Keywords: loan repayment, forecasting approach, machine learning, pseudo-gradient procedure, model structure

References:

1. Aivazyan S.A., Bukhshtaber V.M., Enyukov I.S. Prikladnaya statistika: Klassifikatsiya i snizhenie razmernosti [Applied Statistics: Classification and Dimensionality Reduction]. Moscow, Finansy i Statistika Publ., 1989, 607 p.
2. Vasil'ev N.P., Egorov A.A. [Experience of logistic regression parameters calculation by Newton-Rafson method to estimation resistance to cold of plants]. Matematicheskaya biologiya i bioinformatika = Mathematical Biology and Bioinformatics, 2011, vol. 6, no. 2, pp. 190–199. (In Russ.)
3. Bidyuk P.I., Terent'yev A.N., Gasanov A.S. Construction and Methods of Learning of Bayesian Networks. Cybernetic and System Analysis, 2005, vol. 41, no. 4, pp. 587–598.
4. Sorokin A.S. [Building of scoring cards using logistic regression]. Naukovedenie, 2014, iss. 2. (In Russ.) Available at: Link
5. Wasserman P. Neirokomp'yuternaya tekhnika: teoriya i praktika [Neural Computer Technics: Theory and Practice]. Moscow, Mir Publ., 1992, 240 p.
6. Yakupov A.I. [The decision tree's adaptation for modeling of commercial bank clients solvency]. Iskusstvennyi intellekt = Artificial Intelligence, 2008, no. 4, pp. 208–213. (In Russ.)
7. Merkov A.B. Raspoznavanie obrazov: vvedenie v metody statisticheskogo obucheniya: monografiya [Image recognition: An introduction into statistical learning methods: a monograph]. Moscow, Editorial URSS Publ., 2010, 254 p. (In Russ.)
8. Breiman L. Random Forests. Machine Learning, 2001, vol. 45, iss. 1, pp. 5–32. doi: 10.1023/A:1010933404324
9. Klyachkin V.N., Shunina Yu.S. [System for borrowers' creditworthiness assessment and repayment of loans forecasting]. Vestnik komp'yuternykh i informatsionnykh tekhnologii = Herald of Computer and Information Technologies. Scientific, Technical and Production Monthly Journal, 2015, no. 11, pp. 45–51. (In Russ.)
10. Breiman L. Bagging Predictors. Machine Learning, 1996, vol. 24, iss. 2, pp. 123–140. doi: 10.1023/A:1018054314350
11. Ghodselahi A., Amirmadhi A. Application of Artificial Intelligence Techniques for Credit Risk Evaluation. International Journal of Modeling and Optimization, 2011, vol. 1, iss. 3, pp. 243–249.
12. Krasheninnikov V. R., Klyachkin V. N., Shunina Yu. S. [Pseudo-gradient Procedure as a Method of aggregated Classifiers Updating]. Vestnik komp'yuternykh i informatsionnykh tekhnologii = Herald of Computer and Information Technologies. Scientific, Technical and Production Monthly Journal, 2016, no. 10, pp. 36–40. (In Russ.)
13. Vasil'ev K.K., Krasheninnikov V.R. Statisticheskii analiz izobrazhenii: monografiya [Statistic analysis of images: a monograph]. Ulyanovsk, Ulyanovsk State Technical University Publ., 2014, 214 p.
14. Montgomery D.C. A Note on Forecasting with Adaptive Filtering. Operational Research Quarterly, 1977, vol. 28, no. 11, pp. 87–91.
15. Polyak B.T., Tsypkin Ya.Z. [Pseudogradient adaptation and training algorithms]. Avtomatika i telemekhanika = Automation and Remote Control, 1973, no. 3, pp. 45–68. (In Russ.)
16. Box G., Jenkins G.M., Reinsel G. Analiz vremennykh ryadov. Prognoz i upravlenie [Time Series Analysis: Forecasting & Control]. Moscow, Mir Publ., 1974, 242 p.
17. Lewis C.D. Metody prognozirovaniya ekonomicheskikh pokazatelei [Industrial and Business Forecasting Methods]. Moscow, Finansy i Statistika Publ., 1986, 133 p.

View all articles of issue

ISSN 2311-8768 (Online)
ISSN 2073-4484 (Print)

Journal current issue

Vol. 17, Iss. 1 March 2024

Archive