COMPARISON OF DOUBLE EXPONENTIAL SMOOTHING AND FUZZY TIME SERIES MARKOV CHAIN IN FORECASTING FOREIGN TOURIST ARRIVALS

Abstract

Foreign tourist arrivals are one of the factors that make a positive contribution to a country's economy, especially the addition of foreign exchange. This activity is important for the tourism industry and the government to make policies for progress in the tourism sector. This research aims to forecast data on foreign tourist arrivals, especially land routes. This data set, which is a monthly time series covering the period from January 2018 to October 2023, is sourced from the Central Statistics Agency (BPS). The DES technique is a method that quickly adapts to changes in data patterns and can lessen the impacts of random fluctuations, resulting in more stable estimates. Meanwhile, the FTS-MC approach can handle large data variations by utilizing fuzzy sets. Furthermore, combining fuzzy time series with Markov Chains increases forecast accuracy by taking into account state transitions and probability. The research demonstrates that the DES method produces the MAPE value of 0.108530 or 10% which is obtained from the alpha value of 0.9 and beta 0.2. The MAPE 0.108530 means that the ability of the forecasting model is classified as a good category. In the FTS-MC method, the forecast data is close to the actual data. This is indicated by the MAPE value obtained of 0.086850 or 8%, which means that the ability of the forecasting model is very good. Based on the analysis of the two methods, it is concluded that the FTS-MC method is better applied to data on land-based foreign tourist arrivals.