Modelling of Credit Risk: Random Forests versus Cox Proportional Hazard Regression
American Journal of Theoretical and Applied Statistics
Volume 4, Issue 4, July 2015, Pages: 247-253
Received: May 20, 2015;
Accepted: May 26, 2015;
Published: Jun. 2, 2015
Views 6265 Downloads 318
Dyana Kwamboka Mageto, Jomo Kenyatta University of Agriculture and Technology, Department of Statistics and Actuarial Science, Nairobi, Kenya
Samuel Musili Mwalili, Jomo Kenyatta University of Agriculture and Technology, Department of Statistics and Actuarial Science, Nairobi, Kenya
Anthony Gichuhi Waititu, Jomo Kenyatta University of Agriculture and Technology, Department of Statistics and Actuarial Science, Nairobi, Kenya
In survival analysis several regression modeling strategies can be applied to predict the risk of future events. Often, however, the default choice of analysis tends to rely on Cox regression modeling due to its convenience. Extensions of the random forest approach to survival analysis provide an alternative way to build a risk prediction model. This paper discusses the two approaches in reference to credit management and compares the impact and results of both methods. The Cox Proportional Hazard model displayed a better performance than that of Random Survival Forest when estimating credit risk.
Dyana Kwamboka Mageto,
Samuel Musili Mwalili,
Anthony Gichuhi Waititu,
Modelling of Credit Risk: Random Forests versus Cox Proportional Hazard Regression, American Journal of Theoretical and Applied Statistics.
Vol. 4, No. 4,
2015, pp. 247-253.
Anderssen, P. K. Borgann, Gill keiding N, (1993). Statistical Models Based on Counting Process. Springer series in statistics. New york.
Basel (2000). Principles for the Management of Credit Risk, Basel Commettee. September 2000 1-30.
Breiman, L. (1999). Using Adaptive Bagging to Debias Regression, Technocal report. 547, statistical Depertment UCB.
Creamer, G. (2012). Using Random Forests and Logistic Regression for Performance Prediction of Latin American ADRS and Banks. Journal of Centrum Cathedra, 24-36
Harell, F. E. Califf, R. M. Pryor, D. M. Lee, K. L. Rosati, R. A. Evaluating the Yield of Medical Tests. JAMA. 1982; 247: 2543-2587
Katten, M. Hess, K. and Beck, J. (1998) Experiments to Determine whether Reccursive Partittioning (CART) or an Artificial Neural Network Overcomes Theoretical Limitations of Cox PH Regression, Computer Biomedical Research, 363-373. IBM, Commuter Survey (2014).
May, M. Royston, P. Egger, M. Justice, A. C. and Sterne, J.A.C. (2004) Development and Validation of Prognosis Model for Survival Time Data: Application to Prognosis of HIV Positive Patients Treated with Anti-retroviral Therapy. Statistics Medicine, 23: 2373-2398.
Narain, B. 1992. Survival analysis and the credit granting decision. L. C. Thomas, J. N. Crook, D. B. Edelman, eds.Credit Scoring and Credit Control. OUP, Oxford, U.K., 109–121.
Njanike, K. (2009). The Impact of Effective Credit Risk Mangement on Bank Survival. Annals of the University of Petrosani Economics, 9(2), 173-184.
Wekesa O. (2012), Modelling Credit Risk for Personal Loans Using Product-Limit Estimator. International Journal of Financial Research. (3) 22-32.
Thomas, L. C. Banasik, J. N. Crook. (1999). Not if but when Loans Default. J.Operations Research Society. 50 1185-1190.
Zhang, A. (2009). Statistical Methods in Credit Risk Modelling, University of Michigan, 3 4-27.
Zhou, L. and Wang, H. (2012). Loan Default Prediction on Large Imbalanced Data Using Random Forests, Telkonmika Indonesian Journal of Electrical Engeneering. (10) 1519-1525.