Spatiotemporal Prediction Method Of Urban Management Cases Based On Adaptive Kernel Density Estimation

The vigorous "growth" of the urbanization process has not only promoted the "new normal" of economic development,but also increased the burden of urban management.At present,there are many urban management incidents in cities,such as illegal advertising,road occupation and garbage stacking,which gradually form a "chronic disease" of urban development and reform.Urban fine management is a way of urban management with clear responsibilities and rights.Restricted by many factors such as the rapid development of urban scale,the rapid growth of population and the change of population structure,there are still some problems that are difficult to solve in today's urban management.However,the rapid development of information technology provides support for capturing and detecting laws and patterns in urban management.In particular,spatiotemporal information technology is not only in quantitative analysis,but also in exploring the laws and patterns of urban management in the two dimensions of space and time,providing scientific decisions for the healthy operation and sustainable development of urban planning and urban operation management.This article uses the urban management business data of Xicheng District,Beijing as the data source,and the spatiotemporal prediction model as the research method.A locally adaptive spatiotemporal prediction analysis method and process for urban management cases for hotspot prediction drawing and evaluation is proposed.The main research contents of this paper are as follows.(1)A spatiotemporal prediction model for urban management case data is proposed.In order to be able to respond to urban management case problems in a timely and rapid manner,accurate predictions can be achieved with high spatial and temporal resolution.Aiming at the problem of selecting the optimal bandwidth for kernel density estimation,this paper proposes a locally adaptive spatiotemporal kernel estimation model based on the existing kernel density estimation methods.The model can improve the prediction accuracy according to the temporal and spatial characteristics of urban management cases.First,a new method for estimating spatiotemporal kernel density is proposed,which adds a time dimension to the existing model.Secondly,combine data-driven methods to achieve local adaptation of bandwidth.Finally,with the help of the commonly used evaluation indicators of crime prediction,PAI index and hit rate index,the prediction results of different models are evaluated to verify the calculation efficiency and accuracy of the kernel density estimation in this paper.(2)Propose the analysis framework of time-space prediction research on urban management cases.Temporal and spatial hotspots forecast plays a vital role in achieving efficient and agile,precise grid city management.This paper selects the nonparametric estimation method that only studies the data from the experimental data itself.However,the current popular methods such as kernel density estimation do not take into account the time attribute of spatiotemporal data.Therefore,this paper proposes a spatiotemporal framework for drawing and evaluating prediction hotspots,aiming to explore a prediction model suitable for grid city management problems,so as to study a prediction framework with fast calculation speed,high accuracy and strong practicability.(3)Case analysis of spatiotemporal prediction of urban management cases.Taking the Xicheng District of Beijing as an example,a case study was carried out.The improved spatiotemporal nuclear density estimation method,Pro Map method and traditional nuclear density estimation method were used to construct models and forecasts of the typical urban management case data in 2008.The prediction results of the above model are compared with the data distribution of real cases to realize the accuracy evaluation of the prediction results.The results show that although the three methods can effectively identify the high-risk areas in the hot spots of urban management cases,the hot spots and case hit rates predicted by the spatiotemporal kernel density estimation optimization method proposed in this paper are higher than those predicted by the other two models.The method proposed in this paper improves the computational efficiency.The model in this paper can effectively deal with the spatiotemporal data of urban management cases,and realize the fitting and prediction of spatiotemporal trends,so as to provide reliable technical support for urban management departments to optimize resources and personnel allocation.