长江流域资源与环境 >> 2025, Vol. 34 >> Issue (3): 585-599.doi: 10.11870/cjlyzyyhj202503010

• 自然资源 • 上一篇    下一篇

融合XGBoost-SHAP的重庆市乡村生态旅游资源竞争力测度研究

张慧玲1 ,张虹1* ,孙德亮2   

  1. (1. 重庆师范大学地理信息系统应用研究重庆市高校重点实验室 重庆401331;2. 重庆师范大学地理与旅游学院,重庆401331)
  • 出版日期:2025-03-20 发布日期:2025-03-20

Competitiveness Measurement of Rural Ecotourism Resources in Chongqing with Integrating XGBoost-SHAP

ZHANG Hui-ling1,ZHANG Hong1,SUN De-liang2   

  1. (1.Key Laboratory of GIS Application and Research, Chongqing Normal University, Chongqing 401331, China;2.College of Geography and Tourism, Chongqing Normal University, Chongqing 401331, China)
  • Online:2025-03-20 Published:2025-03-20

摘要: 乡村生态旅游资源竞争力是判别旅游发展潜力的重要依据,对乡村产业振兴有重要意义。XGBoost-SHAP解释性机器学习为乡村生态旅游资源竞争力测度提供可视化的智能工具。融合XGBoost-SHAP构建可解释的机器学习模型,以乡村生态旅游资源点为样本,从自然人文环境、生态资源和旅游基础设施3个维度选择测度指标,以识别重庆市乡村生态旅游资源竞争力水平。结果表明:(1)XGBoost通过学习样本数据潜在模式或规律,高效的识别了乡村生态旅游地的不同竞争力水平,并实现了较高的精度。(2)SHAP提高了XGBoost模型预测的透明度,能识别乡村生态旅游资源竞争力的主导因子,经济活动强度、NDVI、高程和生境质量是对重庆市乡村生态旅游资源竞争力最重要的4个因子,同时也是乡村生态旅游资源开发与可持续利用的主要考量因素。(3)重庆市乡村生态旅游资源强竞争力区域集中在市场、知名景区和交通沿线附近,形成了重庆市西部市场依托、东南和东北部为交通依托和景区依托3种乡村生态旅游发展模式。(4)资源竞争力具有空间溢出效应,距离核心景区远近是影响其强弱的主要因素,旅游基础设施和市场等人文条件也对其产生重要影响。最后,基于研究结论,对平衡重庆市旅游资源的开发与利用,优化旅游空间布局,以及推动“和美乡村”建设方面提供建议。可解释性机器学习模型能快速、准确测度区域乡村生态旅游资源竞争力,可以为旅游资源识别和定量分析提供方法借鉴。

Abstract: The competitiveness of rural ecological tourism resources is an important basis for judging the potential for tourism development and is of great significance for the revitalization of rural industries.XGBoost-SHAP interpretable machine learning provides a visual intelligent tool for measuring the competitiveness of rural ecological tourism resources. In this study, XGBoost-SHAP was integrated to construct an interpretable machine learning model with rural ecotourism resource sites as sample data. Measurement indicators were selected from three dimensions: natural and humanistic environment, ecological resources and tourism infrastructure, in order to identify the competitiveness level of rural ecotourism resources in Chongqing Municipality. The results indicated that: (1) XGBoost, by learning the potential patterns or laws of the sample data, efficiently identified the different competitiveness levels of the rural ecotourism sites and achieved a high accuracy. (2) SHAP improved the transparency of XGBoost, model prediction and was able to identify the dominant factors of competitiveness of rural ecotourism resources. Economic activity intensity, NDVI, elevation and habitat quality were the four most important factors for the competitiveness of rural ecotourism resources in Chongqing Municipality, and they were also the main cooking factors for the development and sustainable utilization of rural ecotourism resources. (3) In Chongqing Municipality, the strong competitiveness of rural ecotourism resources was concentrated in the market, well-known scenic spots and along the transportation route, forming three rural ecotourism development modes: market-dependent in the west of Chongqing Municipality, transportation-dependent in the southeast and northeast, and scenic spot-dependent. (4) There was a spatial spillover effect on the competitiveness of resources, and the proximity to the core scenic area was the main factor affecting its strength. The human conditions such as tourism infrastructure and market also had an important impact. Based on the research findings, suggestions were provided for balancing the development and utilization of tourism resources in Chongqing, optimizing the spatial layout of tourism, and promoting the construction of "beautiful villages." Interpretable machine learning models were proved to be able to quickly and accurately measure the competitiveness of rural ecological tourism resources in the region, which provided a methodological reference for identifying and quantitatively analyzing tourism resources.

No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!