Tourism has a strategic role in the developing of regional economy and society. Surabaya has some object and tourist attractions potentially attracting foreign and domestic tourists to come to Surabaya. The monthly number of tourist who visits to several tourist attractions in Surabaya shows a pattern as time serially data, seasonal, and could be correlated among them. The traditional vector autoregressive (VAR) would be firstly applied to the data. For the comparison, neural networks (NN) couple with VAR structure as an input is proposed to model this data. This paper shows that this proposed method gives better performance than when the data was directly modeled using VAR only. This research also shows that the increasing number of neurons in the hidden layer does not always give effect to the decreasing the value of MAPE as a tool to differentiate the models.
|Number of pages||8|
|Journal||Journal of Theoretical and Applied Information Technology|
|Publication status||Published - 30 Nov 2016|
- Neural network
- Time series
- Tourist visit