Forecasting Tourist Arrivals with Partial Time Series Data Using Long-Short Term Memory (LSTM)

Abstract

Tourism is a source of foreign exchange income, especially in the economic field. Increasing foreign tourist arrivals is essential in supporting the economy of Indonesia. The development of the tourism industry can be seen from the increase in tourist visits every year. Based on data obtained by the Indonesian Central Statistics Agency (BPS), there was an increase and decrease in the number of tourist visits during 2006-2019. Along with these conditions, the provision of various tourism products and services needed to support the industry must be adjusted to prevent financial losses. Unfortunately, tourism products are generally easily damaged, so it is necessary to forecast tourist arrivals. This study aims to predict the arrival of foreign tourists to Indonesia using the Long Short Term Memory (LSTM) method. This method is suitable for sequential data such as tourist arrival data. This shows the results of the evaluation using the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) of 591897.46 for RMSE and 636.2 for MAPE. Based on this research, it can be concluded that LSTM is suitable to be used as a model to predict the arrival of foreign tourists in Indonesia.