—The utilization of artificial neural networks (ANN)
in Islamic banking research is rarely reported. Therefore, this
paper aims to examine the possibility of ANN utilization in case
of predicting mudharabah time deposit return. This paper
compares the accuracy performance of artificial neural
networks (ANN), multiple linear regressions (MLR) and
generalized autoregressive conditional heteroscedasticity
(GARCH) model. Ten years monthly data of six
macroeconomic variables are selected as independent variables.
Meanwhile, the average rate of return of one month
mudharabah time deposit of Indonesian Islamic banks (RR) is
selected as dependent variable. For this purpose, the research
employs Alyuda neuro intelligent software version 2.2 to
develop ANN model and Eviews software version 5.0 to develop
MLR and GARCH model. The performance is evaluated using
visual methodology by analyzing predicted graph and statistical
parameters such as R2, Akaike's information criterion (AIC),
mean absolute error (MAE) and mean absolute standard error
(MASE). Accordingly, this research found that ANN
outperforms MLR and GARCH model in explaining the
volatility of RR. Even though GARCH model outperforms ANN
in making out of sample data prediction, ANN achieves better
accuracy performance in predicting one and two month ahead
of out of sample data. All evidences demonstrate that ANN
model provides more accurate prediction and is appropriate to
be used in Islamic banking research.
—Islamic bank, rate of return, macroeconomic
variables, artificial neural networks, multiple linear regression.
Cite:Saiful Anwar and Yoshiki Mikami, "Comparing Accuracy Performance of ANN, MLR, and GARCH Model in Predicting Time Deposit Return of Islamic Bank," International Journal of Trade, Economics and Finance vol.2, no.1, pp. 44-51, 2011.