Design And Implementation Of Urban Interest Point Mining And Visualization System Based On Density Clustering

With the development of modern society,there are more and more kinds of buildings in the city.These buildings can be collectively referred to as POI(point of interest).There are many practical information in these various kinds of urban interest points,such as the hot spot information of the city interest points,the distribution information of the point of interest point around the fixed-point and so on.How to excavate the above information from the distribution of interest points in cities is of great significance for urban construction departments to make urban planning and businesses to locate reasonable shops.In this paper,the distribution information of city interest points was analyzed by spatial density clustering algorithm DBSCAN and spatial index encoding technology GeoHash,This paper designs and implements a system with the function of mining the distribution of hot spot from the urban interest points and the analysis function of the surrounding interest points,this system has been applied in practice.The main work of the this paper is listed as follows:(1)Data cleaning and data warehousing are carried out for the original POI data.The wrong data and redundant data has been removed.Add the the latitude and longitude information for those data which information are lost according to the address information of these data,the city administrative regions code is unified standard,generate the city administrative region code by the address of interest point,Save all data in a province partition to the database.(2)An improved DBSCAN algorithm is presented.Because the DBSCAN algorithm uses the global parameters Eps and MinPts,it gets poor clustering results on multi-density datasets.The new algorithm used a new distance measuring method called nearest neighbor density distance which makes the new algorithm can clustering properly in multi-density datasets.By analyzing the relationship between the threshold of nearest neighbor density distance and the threshold of nearest neighbor collection,the correlation parameter is determined by the concept of the nearest neighbor density distance change rate and reducing parameter sensitivity.The experimental results show that the improved algorithm is reasonable.(3)An optimized GeoHash peripheral are retrieval algorighm is proposed.In order to accelerate the component analysis module of interest points,encoding the location information of city interest points with spatial index encoding technology GeoHash,the retrieval efficiency is improved.Besides,to solve the problem of loss statistics information,analyzing the encoding rule of the original spatial index encoding technique and proposing a way to optimize the search of surrounding grids.The experimental results show that this method can improve the statistical efficiency and avoid the missing points of interest points.(4)The city point of interest mining system based on spatial density clustering algorithm is designed and implemented,integrate the improved DBSCAN algorithm into the system and finish the hot spot distribution discovery function of the system.With the support of the optimized GeoHash algorithm,improved the response efficiency of the peripheral interest points,and completed the complete function of the whole system.The application results of the actual scene show that the function of the urban interest point mining and visualization system module based on density clustering is effective and reliable,and has two core functions of hot spot discovery and surrounding composition analysis.The system is of great significance to the development of urban construction work and shop location.