Thesis (MCIS - Computer and Information Sciences) -- AUT University, 2007. / Includes bibliographical references. Also held in print (x, 102 leaves : ill. ; 30 cm.) in City Campus Theses Collection (T 332.7 OET)

Thesis (Ph. D.)--Open University. BLDSC no. DX184766.

Zugl.: Dresden, Techn. Universiẗat, Diss., 2007.

After the fifteen years over banking, the financial tsunami reveals that Taiwan banking industry has been in a predicament. In the recent ten years, due to being impacted by the risks of the enterprise finance, the retailer finance, the overseas investment, and even the whole economic system, the banks in Taiwan not only have lost seriously, but also been managed more hardly. How to find out a profit model based on the security, the benefit, and the public welfare principles is the critical issue. The traditional loan to small and medium-sized enterprises that brings the reasonable interest gains and the overall financial intercourse spin-off benefits has become the focal point once again. In order to create the real profit, it is important to control credit risks and the cost of operation. At this time, the government implements the new Basel¢º supervisory standard with the purpose of encouraging the banks to adopt the IRB to estimate the loan credit risks. It has to meet various the lowest operational requirements and statistical analysis patterns as well as should be practiced in the banking loan business definitely. Consequently, to build an internal loan credit scoring system with the scientific method is a key point. The research aims to produce the credit scoring model using a series of logical processes, which derived from the 2,517 small business loan samples from May, 2005 to May, 2006 of one Taiwan commercial bank. It adopted WOE model to evaluate a variety of variables, and sift out the 11 representative and the statistical items. Then, following IRB standard, Logistics Regression and related statistic analysis techniques established the credit estimating method and the linked addition scoring card. Finally, the investigation employed the violation rate distribution, Lorenz¡¦s Curve, K-S Test and Log Odds to make sure the rationality and reliability. Based on this model, there are eight essential variables that affects the verification of the loan to small businesses, including customer present loan situation, the urgent of increasing the loan, repayments custom, and so on, which conform to the banking practical know-how. Therefore, the model could assist the banking employees to calculate the loan credit grades efficiently and further make the accurate judgment.

Konstanz, Univ., Diss., 2008.

Credit scoring is a mechanism used to quantify the risk factors relevant for an obligor’s ability and willingness to pay. Credit scoring has become the norm in modern banking, due to the large number of applications received on a daily basis and the increased regulatory requirements for banks. In this study, the concept and application of credit scoring in a South African banking environment is explained, with reference to the International Bank of Settlement’s regulations and requirements. The steps necessary to develop a credit scoring model is looked at with focus on the credit risk context, but not restricted to it. Applications of the concept for the whole life cycle of a product are mentioned. The statistics behind credit scoring is also explained, with particular emphasis on logistic regression. Linear regression and its assumptions are first shown, to demonstrate why it cannot be used for a credit scoring model. Simple logistic regression is first shown before it is expanded to a multivariate view. Due to the large number of variables available for credit scoring models provided by credit bureaus, techniques for reducing the number of variables included for modeling purposes is shown, with reference to specific credit scoring notions. Stepwise and best subset logistic regression methodologies are also discussed with mention to a study on determining the best significance level for forward stepwise logistic regression. Multinomial and ordinal logistic regression is briefly looked at to illustrate how binary logistic regression can be expanded to model scenarios with more than two possible outcomes, whether on a nominal or ordinal scale. As logistic regression is not the only method used in credit scoring, other methods will also be noted, but not in extensive detail. The study ends with a practical application of logistic regression for a credit scoring model on data from a South African bank. Copyright / Dissertation (MSc)--University of Pretoria, 2010. / Mathematics and Applied Mathematics / unrestricted

Mathematical programming (MP) can be used for developing classification models for the two–group classification problem. An MP model can be used to generate a discriminant function that separates the observations in a training sample of known group membership into the specified groups optimally in terms of a group separation criterion. The simplest models for MP discriminant analysis are linear programming models in which the group separation measure is generally based on the deviations of misclassified observations from the discriminant function. MP discriminant analysis models have been tested extensively over the last 30 years in developing classifiers for the two–group classification problem. However, in the comparative studies that have included MP models for classifier development, the MP discriminant analysis models either lack appropriate normalisation constraints or they do not use the proper data transformation. In addition, these studies have generally been based on relatively small datasets. This thesis investigates the development of MP discriminant analysis models that incorporate appropriate normalisation constraints and data transformations. These MP models are tested on binary classification problems, with an emphasis on credit scoring problems, particularly application scoring, i.e. a two–group classification problem concerned with distinguishing between good and bad applicants for credit based on information from application forms and other relevant data. The performance of these MP models is compared with the performance of statistical techniques and machine learning methods and it is shown that MP discriminant analysis models can be useful tools for developing classifiers. Another topic covered in this thesis is feature selection. In order to make classification models easier to understand, it is desirable to develop parsimonious classification models with a limited number of features. Features should ideally be selected based on their impact on classification accuracy. Although MP discriminant analysis models can be extended for feature selection based on classification accuracy, there are computational difficulties in applying these models to large datasets. A new MP heuristic for selecting features is suggested based on a feature selection MP discriminant analysis model in which maximisation of classification accuracy is the objective. The results of the heuristic are promising in comparison with other feature selection methods. Classifiers should ideally be developed from datasets with approximately the same number of observations in each class, but in practice classifiers must often be developed from imbalanced datasets. New MP formulations are proposed to overcome the difficulties associated with generating discriminant functions from imbalanced datasets. These formulations are tested using datasets from financial institutions and the performance of the MP-generated classifiers is compared with classifiers generated by other methods. Finally, the ordinal classification problem is considered. MP methods for the ordinal classification problem are outlined and a new MP formulation is tested on a small dataset.

Memoria (licenciado en ciencias jurídicas y sociales) / El presente trabajo, describiendo y considerando la situación en que se encuentran las pequeñas y microempresas en nuestro país, analiza sus escenarios en relación al crédito bancario formal. Destacando que los problemas se dan en materia de acceso, monto y plazos. Al mismo tiempo, diferencia los conceptos de crédito a la microempresa y microcrédito, atendiendo al origen de este último, para que el tema no se preste a confusión al momento de abordar una posible solución para hacer frente a los elevados costos de transacción y al riesgo crediticio del cual son presa, no sólo la microempresas, sino también las pequeñas, en el marco del crédito bancario tradicional. Por último, se plantea al Credit Scoring, sistema de evaluación estadístico cuantitativo y luego de analizarlo pormenorizadamente, como una de las mediadas a adoptar por parte de los privados – la banca formal – con el objeto superar la brecha que genera la falta de historial crediticio en este sector empresarial.

This thesis investigates the use of newly-developed survival analysis tools for credit scoring. Credit scoring techniques are currently used by financial institutions to estimate the probability of a customer defaulting on a loan by a predetermined time in the future. While a number of classification techniques are currently used, banks are now becoming more concerned with estimating the lifetime of the loan rather than just the probability of default. Difficulties arise when using standard statistical techniques due to the presence of censoring in the data. Survival analysis, originating from medical and engineering fields, is an area of statistics that typically deals with censored lifetime data. The theoretical developments in this thesis revolve around linear regression for censored data, in particular the Buckley-James method. The Buckley-James method is analogous to linear regression and gives estimates of the mean expected lifetime given a set of explanato ry variables. The first development is a measure of fit for censored regression, similar to the classical r-squared of linear regression. Next, the variable-reduction technique of stepwise selection is extended to the Buckley-James method. For the last development, the Buckley-James algorithm is altered to incorporate non-linear regression methods such as neural networks and Multivariate Adaptive Regression Splines (MARS). MARS shows promise in terms of predictive power and interpretability in both simulation and empirical studies. The practical section of the thesis involves using the new techniques to predict the time to default and time to repayment of unsecured personal loans from a database obtained from a major Australian bank. The analyses are unique, being the first published work on applying Buckley-James and related methods to a large-scale financial database.

Most lending institutions are aware of the importance of having a well-performing credit scoring model or scorecard and know that, in order to remain competitive in the credit industry, it is necessary to continuously improve their scorecards. This is because better scorecards result in substantial monetary savings that can be stated in terms of millions of dollars. Thus, there has been increasing interest in the application of new classifiers in credit scoring from both practitioners and researchers in the last few decades. Most of the recent work in this field has focused on the use of new and innovative techniques to classify applicants as either 'credit-worthy' or 'non-credit-worthy', with the aim of improving scorecard performance. In this thesis, we investigate the suitability of intelligent systems techniques for credit scoring. In particular, intelligent systems that use immunological metaphors are examined and used to build a learning and evolutionary classification algorithm. Our model, named Simple Artificial Immune System (SAIS), is based on the concepts of the natural immune system. The model uses applicants' credit details to classify them as either 'credit-worthy' or 'non-credit-worthy'. As part of the model development, we also investigate several techniques for selecting variables from the applicants' credit details. Variable selection is important as choosing the best set of variables can have a significant effect on the performance of scorecards. Interestingly, our results demonstrate that the traditional stepwise regression variable selection technique seems to perform better than many of the more recent techniques. A further contribution offered by this thesis is a detailed description of the scorecard development process. A detailed explanation of this process is not readily available in the literature and our description of the process is based on our own experiences and discussions with industry credit risk practitioners. We evaluate our model using both publicly available datasets as well as a very large set of real-world consumer credit scoring data obtained from a leading Australian bank. The evaluation results reveal that SAIS is a competitive classifier and is appropriate for developing scorecards which require a class decision as an outcome. Another conclusion reached is one confirmed by the existing literature, that even though more sophisticated scorecard development techniques, including SAIS, perform well compared to the traditional statistical methods, their performances are not statistically significantly different from the statistical methods. As with other intelligent systems techniques, SAIS is not explicitly designed to develop practical scorecards which require the generation of a score that represents the degree of confidence that an applicant will belong to a particular group. However, it is comparable to other intelligent systems techniques which are outperformed by statistical techniques for generating p ractical scorecards. Our final remark on this research is that even though SAIS does not seem to be quite suitable for developing practical scorecards, we still believe that there is room for improvement and that the natural immune system of the body has a number of avenues yet to be explored which could assist with the development of practical scorecards.