Few-shot hotel industry site selection prediction method based on meta learning algorithms and transportation accessibility

Exploring the rationality of hotel location selection is of significant importance for optimizing urban spatial structure and improving tourism service levels. Artificial intelligence provides a data-driven approach for hotel location selection. Therefore, this paper takes the star-rated hotels in the six districts of Tianjin as the research subject and proposes a few-shot hotel location prediction method based on meta-learning algorithms and transportation accessibility. First, the initial location prediction results are obtained through the meta-model. Then, a transportation accessibility calculation model is constructed using spatial syntax for secondary screening. Finally, an appropriateness distribution map is created according to demand levels. The results show that: (1) The meta-model achieves a classification accuracy of 90.45% for star-rated hotels, with a location fitting degree of 91.90%, an improvement of approximately 11% compared to the baseline model; (2) Transportation conditions play a crucial role in the distribution of star-rated hotels, contributing 45% of the classification information; (3) It is recommended that future investments in star-rated hotels focus on areas around Xiaobailou Street, Dawangzhuang Street, and Wudadao Street. Additionally, to verify the overall effectiveness of the model, a comparison with other datasets demonstrates the performance advantages of the meta-model in few-shot location scenarios, providing a practical research path for hotel location decision-making.