Abstract:

In this study, according to the statistics obtained by the Ministry of Culture and Tourism, the estimation of the tourism demand for the whole of Turkey by months is aimed at comparing the predictive accuracy of the methods of "Time Series Analysis" and "Artificial Nervous Networks" to determine the model with high predictive accuracy.             The tourist demand modeling has been designed to identify the appropriate model using the ARMA (Karma Otoregressive Moving Average Process) or ARIMA (Homogenic Non-Standard Processes) methods from time-series techniques.             The purpose of artificial nerve network systems is to determine the high performance model. For this purpose, an extensive number of trials have been trained and conducted to optimal success using different network structures with an algorithm of error retrieval used in the training of artificial nerve networks, advanced powered network structure, consulted learning strategy and structures created with the MLP (Multi-layer Perceptron) model.             As a result of the assessment of the forecast results obtained from the applied methods, it has been found that artificial nerve networks have a higher forecast accuracy according to time series analysis and a method that gives results near real values.

Keywords: