Bank Customer Classification in Indonesia: Logistic Regression Vis-à-vis Artificial Neural Networks

This paper aims to identify factors distinguish Islamic and conventional bank customers in Indonesia. It tries to relate between bank customers’ religiosity, assessment upon certain factors such as bank performance, bank advertisement and main reasons of using banking services towards their decision on which bank they had joined. Logistic regression and neural networks models are used to answer the research questions based on 520 customers reside in Jakarta. Data collection is done through a direct survey using self administered questionnaire. The results from logistic regression and neural networks models demonstrate that shariah compliant issues, customers’ awareness on the fatwa announced by National Ulama Council on the impermissibility of bank interest, safety of fund as main reason of using banking services and customers’ perception on bank advertisement are the significant factors which classify the bank customers in Indonesia. Nonetheless, neural network classifies better than logistic regression.