Τεχνικές εξόρυξης γνώσης στον τραπεζικό τομέα. Μελέτη περίπτωσης: πρόβλεψη πιθανότητας αθέτησης στις πιστωτικές κάρτες
Data mining techniques in the banking sector. Case study: predicting credit card default probability
View/
Abstract
Nowadays, technology is one of the most developed sciences. In recent decades, there is a tendency by term technology to refer mainly to computer technology. In today's reality, the major needs and challenges in this area are both the management of huge volumes of data, which have grown exponentially and continue to grow, and the utilization of this data in an optimal way to produce valuable and reliable data and explore new areas and information yet unknown. The advantages of achieving the above goals are various and can have a catalytic effect on many business sectors. The financial sector is one of the foremost, especially at a time when the global and european economies are in a phase of significant slowdown, while trying to recover from the 2008 global economic downturn. It is therefore easily understood that particularly for the banking sector, which is the heart of the economy, the use of sophisticated technologies can help in automation and optimization of processes and thereby in reducing risk and in the increase of profitability.
The aforementioned areas of interest and the relationship between them are presented and analyzed in this thesis, which aims to capture and combine basic theoretical knowledge on new technologies and innovations and relate them to the banking industry. In particular, the thesis initially (Chapters 1 and 2) addresses the concepts of Big Data, Data Mining, Machine Learning and the use of Artificial Intelligence solutions in the business world. Subsequently (Chapter 3) the author focuses on the relationship between the above technologies and the financial sector, and in particular on their relationship with the assessment and mitigation of one of the major risks facing banks, credit risk. The analysis is progressively moving to a more specialized area, that of assessing the probability of default on the credit card portfolio, a category of loans with particular characteristics. In the next two chapters, machine learning algorithms are applied on an external bank’s data set to predict the probability of default and evaluate the results.
The path which the reader is invited to follow in this particular thesis, in which different - interlinked however - meanings are presented, is a result of both the author's interest in these subject matters and his academic and professional background.