Algorithmic Forecasting of Tourist Mobility: Implementation of Fuzzy Time Series in High-Variability Aviation Data

Abstract

Accurate forecasting of tourist arrivals is a critical determinant for strategic planning and operational efficiency in island destinations. However, forecasting domestic tourist mobility through airport gateways, such as Lombok International Airport (LIA), presents a significant challenge due to the high volatility and non-linear characteristics of aviation data. Conventional statistical models often fail to capture these dynamic fluctuations effectively. To address this issue, this study proposes an algorithmic forecasting framework using the Fuzzy Time Series (FTS) Chen model. The methodology involves processing monthly arrival data through a structured sequence: defining the universe of discourse, partitioning intervals, fuzzification, establishing Fuzzy Logical Relationships (FLRs), and performing defuzzification. The model's performance was rigorously evaluated using the Mean Absolute Percentage Error (MAPE). Empirical results demonstrate that the FTS Chen algorithm is highly effective for stable datasets, achieving a forecasting accuracy with a MAPE as low as 9.23% for foreign tourist arrivals. In contrast, the model exhibited higher error rates for domestic tourist data, attributed to significant seasonal volatility and external shocks. These findings confirm that while the proposed soft computing approach is robust for detecting trends in stable tourism flows, highly fluctuating domestic markets may require hybrid optimization. Practically, this study provides airport authorities with a quantitative tool to anticipate visitor volume and optimize resource allocation in the post-pandemic era.