Forecasting number of tourists with Pi-Sigma artificial neural networks

Artificial neural networks are powerful alternatives to well-known forecasting methods. It is asserted that high order neural networks produce better forecasting results than classical artificial neural networks. In this study, the forecasting performance of pi-sigma artificial neural networks on number of foreign arrivals to Turkey time series are investigated. The performance of pi-sigma artificial neural networks is compared with classical forecasting methods and alternative artificial neural networks by using various statistics and Nemenyi test. The obtained results presented by tables and figures.