A predict-then-optimize framework for customized tourism service pricing and networked multimodal transport

This is the authors' abstract. We don't add key points for this paper.

As attractive tourist spots are often located outside urban areas, tourists usually use networked multimodal transport to reach destinations. However, segment-by-segment travel degrades operational efficiency and passenger comfort. For this tourism routing problem, this study proposes a predict-then-optimize framework. In the prediction phase, Cumulative Prospect Theory (CPT) captures passenger heterogeneity and bounded rationality in travel choices. In the optimization phase, a behavior-driven multi-objective optimization model is built based on predicted demand fluctuations to derive optimal pricing strategies and vehicle allocation schemes. The model minimizes passenger travel costs and maximizes operator revenue subject to such constraints as capacity and fare limits. An adaptive large-neighborhood search algorithm incorporating simulated annealing criteria is developed to solve the model. Results show that the optimized scheme increases operator revenue by 5.17% over traditional segment-by-segment pricing without ancillary travel services and reduces passenger travel costs by 4.26%. These findings verify market-oriented integrated pricing tactics.