Recurrent neural network with Extended Kalman Filter for prediction of the number of tourist arrival in Lombok

Abstract

Tourism has become a major sector for economic development on Lombok. Tourist expenditure in Lombok give a good implications on public revenue. Tourist expenditures are not only distributed to the tourism sector, but also to other sectors. Prediction of tourists visit is very important as an information and planning for the future. Prediction is one of very important element in decision, because the effectiveness of the decision generally depends on several factors in present and past. This research tries to predict tourist arrival by examining time-series data on tourist arrival in Lombok by using Recurrent Neural Network with a training algorithm Extended Kalman Filter. Based on the accuracy of prediction in the data testing, this method are good for predicting time series data. The best result of the prediction in the data testing showed MSE value is 0.052232 with the accuracy of the prediction is 86.059 %.