Analysis Of Spatial Pattern Of Regional Catering Industry And Its Driving Force

In the context of the new era,the Internet economy empowering the catering industry has brought new development opportunities and challenges to the catering industry.As an important part of urban commerce,the catering industry has increasingly become a visual representation of the city’s residential comfort,cultural prosperity,and economic development.Hot pot is an important business in the restaurant industry and is a group participation food culture formed in urban space by the interweaving of community life circle and commercial service circle.As a city with a prosperous hot pot culture and a large number of hot pot restaurants,a study on Chongqing hot pot catering can help optimize the spatial pattern of the city’s catering service industry on the supply side and meet the demand of city residents for good food on the demand side.First,the spatial pattern of hotpot restaurants is analyzed from the global and four town cluster perspectives using standard deviation ellipses,mean nearest neighbor analysis,and kernel density estimation.The spatial distribution of hot pot restaurants in Chongqing is uneven,with the overall spatial distribution characteristics of "more in the west,less in the east,and unique in the center",and the spatial distribution is very uneven.Chongqing hot pot restaurant stores are clustered in Chongqing city and the four major town clusters,and the clustering of hot pot restaurants in the city is higher than the clustering of the four major town clusters.The results of the nuclear density analysis show that the spatial clustering of hot pot restaurant stores in Chongqing is very obvious,with a clear central pointing at the city scale.Second,the independent and interactive influences of the geographic detector quantification factors were used to analyze the key driving factors affecting the spatial pattern of hot pot restaurant stores.It was found that different environmental factors are sensitive to spatial scale,and the independent factor detection results show that the spatial distribution of hot pot restaurants in Chongqing is most significantly influenced by store reputation,followed by the influence of per capita consumption,the number of residential communities,the number of business districts and the number of parking lots,which are all environmental factors that should be focused on when selecting the location layout of hot pot restaurants in Chongqing.The single-factor detection results of the four town clusters have obvious commonalities: the number of communities,per capita consumption level,the number of shopping areas,and the number of attractions are the key factors in the single-factor detection results.Factor detection helps different districts and counties focus on optimizing key environmental variables when making spatial adjustments to hot pot restaurant stores.Finally,we obtain and fuse the multi-source spatial data to establish the site selection feature matrix and build the site selection prediction model.Support vector machine models built with two different kernel functions are first compared,and it is found that the support vector machine model built based on radial basis kernel functions has better performance.Further,the Lightgbm algorithm,RF algorithm,and Rbf SVM algorithm are comprehensively compared through experiments,and the experimental results show that the Lightgbm algorithm works best under 50 times of cross-validation,and the average accuracy is as high as 0.9267,and the model has a high fitting accuracy.The site-selection prediction model combined with GIS spatial analysis was built by Lightgbm algorithm to obtain 578 candidate location points with high suitability for new hot pot restaurants to complete accurate site selection,and the site-selection model has realistic reference significance.