Processing of large amounts of data on a credit scoring example using neural network technology

Abstract

Nowadays there is the growing problem of mining large amounts of data. This article is dedicated to the issue of development of a credit scoring model as an example of processing large volumes of data. Some data mining algorithms are described in the paper. Three methods have been used for the experiment; namely, logistic regression, decision trees and neural networks. All of them have been applied for the modeling of credit scoring. According to the results of a comparative analysis, neural networks have been selected as a technology for the credit scoring model design. The main aim is to choose the best method of data mining and construction of predictive credit scoring without using expensive software, together with the ability for self-learning and updating. To implement and achieve the goal, the following tasks have been undertaken: collecting and preparing the initial data, analysis and selection of available technologies and methods of solution, to determine the most suitable method of data mining to build a credit scoring system, and now the project is on the way to creating an expert system.