Indian Journal of Science and Technology

Abstract

Background: Tourism industry is very important for a country. Many tourists travel into a country will help and improve its economic growth. Methods: Many researchers used Backpropagation Neural Network (BPNN) for predicting tourist arrivals in a country. As the result, BPNN is proven to give good results, but the accuracy is still less than optimal. This study uses series dataset from the arrival of foreign tourists in the district of Central Java’s: Magelang, Solo, and Wonosobo from 1991 to 2013. We compared the performance of BPNN, K-Nearest Neighbor (KNN) and Multiple Linier Regression (MLR). Genetic Algorithm is used to optimize the parameters of BPNN, such as learning rate, training cycle, and momentum. The performance is measured by Root Mean Square Error (RMSE). Findings: BPNN produces small error of prediction compare to KNN and MLR. KNN performed the worst when used to predict. Improvements: Genetic algorithm proved to be able to optimize the parameters of BPNN. GA is able to minimize the error of the prediction of BPNN.

 Keywords: Data Mining Algorithm, Forecasting, Genetic Algorithm, Neural Network, Tourism Arrival