Stanovení pravděpodobnosti defaultu klientů pomocí scóringového modelu

Abstract

The Master thesis is focused on the determination of the probability of clients’ default using scoring model. The objective of the thesis is to create scoring model to determine the probability of clients’ default using the binary logistic regression followed by verification of the model. The thesis is divided into five main chapters. The thirst one is devoted to introduction and the last one to conclusion. In the second chapter there are described various types of risks. The attention is focused mainly on credit risk as it has the key role in the process of evaluation of clients’ creditworthiness. The third chapter is devoted to the description of scoring and rating models and characterization of different statistical methods that are used for the creation of predictive models. The fourth chapter represents a practical part of the thesis where the predictive scoring model is constructed based on a real dataset of retail loans. The model was set up using binary logistic regression in the statistical program IBM SPSS Statistics 23. In the end of the thesis achieved results are summarized and final model is evaluated.