A planning model for dedicated tourist bus routes based on an improved genetic-greedy algorithm and machine learning

Abstract

Background: This study addresses the challenges posed by the growing number of self-guided tourists and proposes an optimized tourist bus route planning model to enhance visitor satisfaction and support sustainable tourism. Methods: Using machine learning algorithms-adaptive boosting (AdaBoost), support vector machine (SVM), naive Bayes, and K-Nearest Neighbor (KNN)-we analyze sentiment in tourist reviews, with SVM showing the best performance. A multi-criteria evaluation model combining analytic hierarchy process (AHP) and the entropy weight method (EWM) identifies key satisfaction factors, which are integrated into the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) model and the rank-sum ratio (RSR) method to recommend attractions. Results: The optimized route is determined using a modified Genetic-Greedy Algorithm (GGA), which improves convergence speed by 94.489% compared to traditional genetic algorithms. Applied to a case study in Tibet, the model achieved a 94.6% satisfaction rate, demonstrating its effectiveness and adaptability for diverse tourism planning contexts.