A Study of Sustainable Tourism Development Modeling Based on Multi-Objective Optimization and Machine Learning

Abstract

This paper aims to promote sustainable tourism by developing a multi-objective optimization-based sustainable tourism model and a visitor flow balancing model, addressing over-tourism challenges in destinations such as Juneau, Yellowstone National Park, Venice, and the Maldives. The first model integrates ARIMA and EMA for forecasting tourist volumes and environmental indicators, employs multiple linear regression to evaluate resident satisfaction, and incorporates a feedback mechanism linking carbon footprint and tourism revenue. The paper further conducted a sensitivity analysis on tourist volume, carbon tax rate, and resident satisfaction, confirming the robustness and adaptability of the sustainable tourism model. The second model utilizes a probabilistic framework to balance visitor flow, exemplified through 12 attractions in Yellowstone. Seasonal patterns are predicted using the SARIMA model, and a variance-mean ratio is minimized to optimize distribution. The models demonstrate strong scalability and practical value for real-world tourism management.