Optimal Adaptive Neuro-Fuzzy Inference System Architecture for Time Series Forecasting with Calendar Effect

This paper discusses a procedure for model selection in ANFIS time series forecasting with calendar effect. Calendar effect is different from the usual trend and seasonal effects. Therefore, when it occurs, will affect economic activity during that period create new patterns result inaccurate forecasts decision making if not considered. The focus on strategy to find appropriate input variable number of membership functions (MFs) based Lagrange Multiplier (LM) test. ARIMAX stochastic used at preprocessing stage capture variations data. observed Eid al-Fitr holiday Indonesia, country largest Muslim population world. data Tanjung Priok port passengers as case study. shows hybrid ARIMAX-ANFIS LM test can be an effective forecasting. Empirical results show use provides more accurate predictions indicated by smaller RMSE MAPE values than without variable.