Abstract: Studying the hot spot distribution of urban facilities is of great significance to grasp the current urban form. Traditional facility hot spot recognition methods tend to ignore the feature scale of facilities, focus on regional research, and lack a method system for accurately extracting facility hot spots. To solve the above problems, this paper proposed a method of hot spot recognition considering attribute characteristics,and takes the residential facilities in Beijing as an example. Firstly, the attribute values of facilities are used as weights to estimate the density value surface generated by weighted kernel density estimation, and the extreme points are extracted by using the extreme point detection model. Then use Getis-Ord G_i^* statistics for spatial autocorrelation analysis to generate statistically significant hot spots, and select extreme points to obtain hotspots. The experimental analysis shows that the method can accurately and effectively identify the hot spots of the facilities and make a reasonable classification, providing a diversified perspective for the research on the spatial layout of urban facilities.

Key words: kernel density estimation, extreme point, hot spot, attribute characteristic, urban facility