基于DBSCAN算法的民宿集群识别、分布格局及影响因素——以南京市为例  

CLUSTER IDENTIFICATION, DISTRIBUTION PATERN AND INFLUENCING FACTORS OF B & B BASED ON DBSCAN: A CASE STUDY OF NANJING

作　　者：马小宾 侯国林[1,2] 李莉 杨燕 MAXiao-bin;HOU Guo-lin;LI Li;YANG Yan(School of Geography,Nanjing Normal University,Nanjing 210023,China;Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application,Nanjing 210023,China;Zhongbei College,Nanjing Normal University,Nanjing 212300,China)

机构地区：[1]南京师范大学地理科学学院,南京210023 [2]江苏省地理信息资源开发与利用协同创新中心,南京210023 [3]南京师范大学中北学院,南京212300

出　　处：《人文地理》2021年第1期84-93,共10页Human Geography

基　　金：国家自然科学基金项目(41771151);江苏高校哲学社会科学基金项目(2015SJB113)。

摘　　要：以南京市Airbnb网站民宿商家为例,综合利用DBSCAN聚类算法、空间变异函数、地理加权回归模型刻画南京市民宿的空间特征及影响因素。结果表明:(1)南京市民宿已经形成"25124"的5等级14集群格局,呈现"两横两纵"的"井"字形空间布局特征;(2)民宿密度分布存在空间异质性,各个街道之间存在较强的空间自相关,民宿密度分布受到系统自身的影响大于随机性因素的影响;(3)不同因素对民宿空间格局的作用方向和影响程度存在较大异质性,路网密度、公交站点密度、人口密度以及商品房平均价格表现出较强的正相关,距最近地铁站距离、超市商城密度、距4A级以上景区最近距离以及距4星级以上酒店最近距离表现为负相关,不同因素回归系数的分布空间差异显著。Recently, as an emerging accommodation product, B & B has triggered changes in the tourism market. It can fully meet the personalized and diverse needs of tourists and become a new growth pole for tourism destinations. Taking Nanjing as an example, we used crawler software to obtain the data of Airbnb’s B &B, relied on the DBSCAN clustering algorithm to identify the clusters of B & B in Nanjing, and analyzed the spatial distribution characteristics of B & B with the help of the spatial variation function model. Finally, the geographically weighted regression model was used to explore the impact and spatial differences of eight indicator factors in five dimensions, including transportation, commerce, population level, tourism resources,and internal relationships in the industry. The results show that: 1) The agglomeration effect of B & B space in Nanjing is significant, forming a 5-level 14-cluster of "25124" pattern and presenting a spatial layout characteristic of "two horizontal and two vertical" tic-tac-toe shape;2) The density distribution of B & B has a certain spatial heterogeneity, and the nugget coefficient of the spatial variation function reaches 0.093, indicating that there is a strong distribution of B & B spatial autocorrelation. 3) Different factors have great heterogeneity in the direction and influence of the spatial pattern of the B & B.

关 键 词：民宿 DBSCAN聚类算法 空间格局 影响因素 南京市 

分 类 号：K901[历史地理—人文地理学]