Evaluating the borrower's creditworthiness of loans using data mining methods
DOI:
https://doi.org/10.20535/SRIT.2308-8893.2019.2.04Keywords:
decision trees, logistic regression, Bayesian network, fuzzy logic, probability of default, Mamdani conclusionAbstract
The actual task of creditworthiness based on the expert and scoring approach was considered. The analysis of the subject area was performed and the main methods of mathematical modeling and a credit risk assessment were analyzed; mathematical models for analyzing the credit risks of individual borrowers based on alternative methods were proposed; mathematical models have been developed for analyzing the credit risks of individual borrowers based on decision trees, logistic regression, Bayesian networks, and fuzzy logic. It has been found that the model based on fuzzy logic for solving the problem of determining the probability of default for a loan borrower is more accurate, this is indicated by the calculated accuracy of models. This is due to the possibility of using the fuzzy logic method with fuzzy Mamdani’s conclusion to precisely establish the cause-and-effect relationships between the characteristics-factors of the task and their influence on the initial variable.References
Krjuchkovskij V.V. Ekspertnaja sistema otsenki kreditosposobnosti bankovskih klientov na osnove metodov nechetkoj logiki i seti Bajesa / V.V. Krjuchkovskij, S.A. Babichev, A.V. Sharko // Ekonomika nauchno-tehnicheskogo progressa. — 2009. — № 1. — S.197–205.
Allen S. Financial risk management: A practitioner’s guide to managing market and credit risk / S. Allen. — Hoboken, N.J.: John Wiley & Sons, Inc., 2003. — 567 p.
Hennie van Greuning. Analyzing and managing banking risk: a framework for assessing corporate governance and financial risk / Hennie van Greuning, Sonja Bratanovic. — 2nd ed.
Kostjuchenko N.S. Analiz kreditnyh riskov / N.S. Kostjuchenko. — SPb.: ITD "Skifija", 2010. — 440 s.
Bidjuk P.I. Modeli otsinky ryzykiv kredytuvannja fizychnykh osib / P.I. Bidjuk, Ye.O. Matros // Kibernetyka ta obchysljuval'na tekhnika. — 2007. — № 153. — S. 87–95.
Bidjuk P.I. Porivnjal'nyj analiz kharakterystyk modelej otsinjuvannja ryzykiv kredytuvannja / P.I. Bidjuk, N.V. Kuznyetsova // Naukovi visti NTUU "KPI". — 2010. — № 1. — S. 42–53.
Kolpakov V.M. Teorija i praktika prinjatija upravlencheskih reshenij: ucheb. posobie / V.M. Kolpakov. — 2-e izd., pererab. i dop. — K.: MAUP, 2004. — 504 s.
Agresti A. Building and applying logistic regression models. An Introduction to Categorical Data Analysis / A. Agresti. — Hoboken, New Jersey: Wiley, 2007. — 138 p.
Shariff A. The Comparison Logit and Probit Regression Analyses in Research / A. Shariff, A. Zaharim, K. Sopian. — 2009. — Vol. 27, N 4. — P. 548–553.
Terent'ev A.N. Sravnenie metodov intellektual'nogo analiza dannyh pri otsenivanii kreditosposobnosti fizicheskih lits / A.N. Terent'ev, P.I. Bidjuk, A.V. Mironova, N.Ju. Medin // Problemy upravlenija i informatiki. — K.: IKI NANU-NKAU, 2009. — № 5. — S. 141–149.
Terent'ev A.N. Metod verojatnostnogo vyvoda v bajesovskih setjah po obuchajuschim dannym / A.N. Terent'ev, P.I. Bidjuk // Kibernetika i sistemnyj analiz. — 2007. — № 3. — S. 93–99.
Bidjuk P.I. Osnovni etapy pobudovy i pryklady zastosuvannja merezh Bajyesa / P.I. Bidjuk, N.V. Kuznyetsova // Systemni doslidzhennja ta informatsijni tekhnolohiyi. — 2007. — № 4. — S. 26–39.
Heckerman D. Bayesian Networks for Data Mining / D. Heckerman // Data Mining and Knowledge Discovery. — 1997. — № 1. — P. 79–119.
Kuznyetsova N.V. Hibrydni merezhi Bajyesa: osnovni osoblyvosti i tochni metody formuvannja vysnovku / N.V. Kuznyetsova // Pratsi Odes'koho politekhn. un-tu. — Odesa, 2009. — Vyp. 1(31). — S. 114–121.
Kuznyetsova N.V. Systemnyj pidkhid do analizu kredytnykh ryzykiv z vykorystannjam merezh Bajyesa / N.V. Kuznyetsova, P.I. Bidjuk // Naukovi visti NTUU "KPI". — 2008. — № 3. — S. 11–24.
Nedosekin A. Metodologicheskie osnovy modelirovanija finansovoj dejatel'nosti s ispol'zovanie nechetko mnozhestvennyh opisanij / A. Nedosekin. — SPb, 2003. — 280 s.
Zajchenko Ju.P. Otsenka kreditnyh bankovskih riskov s ispol'zovaniem nechetkoj logiki // Systemni doslidzhennja ta informatsijni tekhnolohiyi. — 2010. — № 2. — S. 37–54.
Shovhun N.V. Analiz kredytospromozhnosti pozychal'nyka za dopomohoju metodiv z nechitkoju lohikoju / N.V. Shovhun // Visnyk NTUU "KPI". Informatyka, upravlinnja ta obchysljuval'na tekhnika: zb. nauk. prats'. — K.: Vek+, 2012. — № 55. — S.169–173.
Shovgun Natalia. Fuzzy neural networks for evaluating the creditworthiness of the borrowers / Natalia Shovgun // Information theories & applications. ITHEA IBS ISC. — 2014. — Vol. 21. — P. 54–257.
