Development of Decision Support Model in the Field of Credit Scoring Using Predictive Analytics for a Large Commercial Bank

Today, the banking industry is one of the most competitive sectors. Therefore, banks have to constantly study competitors, as well as conduct in-depth analysis of the target group of existing and potential customers, investigating factors that can impact their core business. In addition, many potential customers who are interested in credit products have unsatisfactory credit history, or do not have credit history at all. There are cases when fraudsters or unscrupulous borrowers use the loans. For these reasons credit scoring saves time, resources and money of a bank. Currently, in the financial market, in addition to the well-known methods of credit scoring, there is an active development of advanced analytics methods, including machine learning on the own data. The huge amount of data accumulated in the financial market motivates banks to develop their own predictive models. The purpose of the master's thesis is to develop the decision support model in credit scoring using predictive analytics for the large commercial Bank. The object of the research is the system of lending in commercial banks, the subject – credit scoring and tools to improve the efficiency of its work. The First Chapter is devoted to the theoretical background of the study. It provides an overview of the banking industry, discusses the use of predictive analytics in banks, describes the lending system, as well as machine learning methods used in credit scoring. The Second Chapter describes the main activities of the commercial Bank. The methodology of data analysis is proposed for use, requirements and assumptions to the decision support model are formulated. The Third Chapter presents the results of the data exploration and forecasting methods, as well as the practical implementation of the predictive model in Anaconda environment (Python) on the example of the commercial Bank data. All the tasks set in the study were successfully solved.