Introduction of scoring and its realization by means of discriminant analysis and neural networks

Abstract

The paper used methods for time series forecasting and risk analysis of the quality of the consumer loan portfolio and compare their quality, used technique of scoring and its implementation through the use of discriminant analysis and neural networks to predict the likelihood of timely repayment of the loan borrower. Forecasting is done in software product «Statistica». In this paper we implemented ARIMA model, seasonal decomposition, exponential smoothing, discriminant analysis and neural network techniques.