Commercial Bank Credit Amount Forecasting using Data Mining Technique

Abstract

The purposes of this research were to develop and compare the effectiveness of the model of forecasting the amount of credit of the commercial banks in Thailand using the data mining techniques by 3 methods, as follows; 1) Linear Regression, 2) Multi-Layer Perceptron and 3) Support Vector Machine for Regression. The data used for this research is related to all factors used for identifying the amount of credit of the commercial banks in Thailand i.e. the amount of deposit of the commercial banks, minimum loan rate, non-performing loan and the amount of credit of the commercial banks in Thailand. The data is from 1 January 2011 to 31 December 2021 and divided into 10 data sets for model creation and 1 data set for model testing. The comparison of the effectiveness for the model of forecasting the amount of credit of the commercial banks that suitable for testing data in 2021 found that the Linear Regression was the highest performance model for forecasting at 2.54% of MMRE and the Support Vector Machine for Regression was at 3.62% of MMRE subsequently.