ESTIMATION OF LOAN APPLICANTS DEFAULT PROBABILITY APPLYING DISCRIMINANT ANALYSIS AND SIMPLE BAYESIAN CLASSIFIER

In commercial banks risk management the credit risk measurement of each client is very important for the ability to discriminate reliable clients from not reliable. The need for models that predict defaults accurately is imperative, because bank crediting clients can take either preventive or corrective action. One of possible quantitative methods for solving credit risk estimation problems is discriminant analysis. In this paper 27 discriminant analysis models of various researchers for classification of companies were analyzed. The average classification accuracy of these models was evaluated. Often discriminant analysis is used as method to classify bank‘s clients into two classes: default and not default. So the discriminant analysis model was developed to classify Lithuanian companies. The best classification accuracy rates were reached by model analyzing data about companies of 2 years. In this research according to 4 financial ratios, 8 ratings scale was created. The simple Bayesian classifier was applied for calculation of posterior probabilities to default of each rating. Created rating scale increased the correct classification rate of model from 84% to 98%.

[1]  Ron S. Kenett,et al.  Statistics for Business and Economics. , 1988 .

[2]  Anthony C. Antonakis,et al.  Assessing naïve Bayes as a method for screening credit applicants , 2009 .

[3]  Peter Carruthers,et al.  The Innate Mind, Volume 3 , 2005 .

[4]  Danuta Zakrzewska.  ON INTEGRATING UNSUPERVISED AND SUPERVISED CLASSIFICATION FOR CREDIT RISK EVALUATION , 2007 .

[5]  S. Stich,et al.  The Innate Mind: Volume 3: Foundations and the Future , 2008 .

[6]  Magdalene Marinaki,et al.  Optimization of nearest neighbor classifiers via metaheuristic algorithms for credit risk assessment , 2008, J. Glob. Optim..

[7]  W. Karwowski.  International encyclopedia of ergonomics and human factors , 2001 .

[8]  Francesco Ciampi,et al.  Using Economic-Financial Ratios for Small Enterprise Default Prediction Modeling: An Empirical Analysis , 2008 .

[9]  Dorota Witkowska.  Discrete Choice Model Application to the Credit Risk Evaluation , 2006 .

[10]  Peter Carruthers,et al.  Foundations and the future , 2007 .

[11]  Hussein A. Abdou.  An evaluation of alternative scoring models in private banking , 2009 .

[12]  Kin Keung Lai,et al.  Neural Network Metalearning for Credit Scoring , 2006, ICIC.

[13]  Kin Keung Lai,et al.  Credit scoring using support vector machines with direct search for parameters selection , 2008, Soft Comput..

[14]  Peter E. Kennedy.  A Guide to Econometrics , 1979 .

[15]  R. Suganya,et al.  Data Mining Concepts and Techniques , 2010 .

[16]  J. Priestley,et al.  ASSESSMENT OF EVALUATION METHODS FOR BINARY CLASSIFICATION MODELING , 2003 .

[17]  Thilo Liebig,et al.  Credit Risk Factor Modeling and the Basel Ii IRB Approach , 2003, SSRN Electronic Journal.

[18]  Juliana Yim,et al.  A comparison of corporate distress prediction models in Brazil: hybrid neural networks, logit models and discriminant analysis , 2005 .

[19]  David Wooff,et al.  Bayes Linear Statistics: Theory and Methods , 2007 .

[20]  Alan H. Fielding,et al.  Cluster and Classification Techniques for the Biosciences , 2006 .

[21]  G. Falavigna.  Models for Default Risk Analysis: Focus on Artificial Neural Networks, Model Comparisons, Hybrid Frameworks , 2006 .

[22]  Arindam Bandyopadhyay,et al.  Predicting probability of default of Indian corporate bonds: logistic and Z‐score model approaches , 2006 .

[23]  E. Altman,et al.  Modelling Credit Risk for SMEs: Evidence from the U.S. Market , 2007 .

[24]  Kuldeep Kumar,et al.  Artificial neural network vs linear discriminant analysis in credit ratings forecast: A comparative study of prediction performances , 2006 .

[25]  The Comparative Analysis of the Models in Default Warning of the Credit Clients in Commercial Banks , 2007 .

[26]  E. Altman,et al.  Modeling Credit Risk for Smes: Evidence from the Us Market , 2005 .