Development of a neuro-fuzzy model for airline passenger forecasting

A five-layer neuro-fuzzy model with Sugeno fuzzy rules is developed to model the dynamics of airline passengers. The effectiveness in modeling, prediction and forecasting is validated by a set of data containing the consumer price index, the exchange rate, the gross national product, and the number of airline passengers traveling abroad from 1995 to 2007. A modified moving average method is applied to predict the input set for the model in forecasting the number of airline passengers. Simulation results show that the neuro-fuzzy model with Sugeno fuzzy rules is effective in prediction and accurate in forecasting. The input error from the modified moving average method is attenuated by the neuro-fuzzy model to yield better forecasting results.