The Design And Implementation Of Data Analysis System Based On Location

Mobile Internet era is the era of location-based large scale data,with more data generated by location positioning,more researches are being conducted on location data.At present,the general approaches to location data are to use the traditional statistical analysis methods and to visualize the analysis results,without satisfying user needs to interact with models;on the other hand,the analysis on location data is limited to the users' target needs,ignoring the attributes like sparsity and imbalance,which are caused by the threshold value on collection time,distribution on map and bias caused by users,ultimately reducing the accuracy of location data analysis.Taking the attributes of time and space in location data as the background,the passage applies time series analysis and location division techniques to solve sparsity,and at the same time,applies pattern division to solve imbalance,and combines these models to satisfy user demands,which impels the visualization and analysis of the location information become the research focus.This paper focuses on the background and requirements of data platform to analyze location data for users,studies needed theoretical knowledge and technical solutions when implementing location-based data analysis,and finally designs and develops a location-based data analysis system.The main research contents are as follows:1.Merge time phasing technique to solve sparsity of data.In the time dimension of location data,this paper applies Bayesian change-point detection to divide time series.In the meantime,associated with basic analysis,the system designs the analysis flow in time series,and achieves the function modules of basic analysis and timing division.2.Introduce pattern division on location data to solve the imbalance.The passage applies relationship management model RFM and clustering model K-means to divide data pattern,for the sake of output what the users need.Besides,the function modules of model increment,pattern selections are realized.3.Realize area division on location data to output what the users demand.Satisfying the requirements on area settings from users,this paper uses DBSCAN clustering model to realize the conversion from the position point data to region data.According to the user needs,the output process of the region area in specific field is designed.Besides,the function modules of region generation and field matching are realized.4.Research on the technique scheme to processize SPARK analysis and visualize location data on large scale data.The interacted models in this article,satisfying user demands,will integrate the scattered techniques in the form of functional modules,and eventually transform into a complete system.At last,the validity of these methods is verified by double case tests.Based on the above research contents and achievements,this paper constructs and implements a location-based data analysis system that aim to solve sparsity and imbalance,satisfy user demands,via time phasing,data division,area division and field match techniques,for the data system to effectively output the location area in specific field.In the process,the system implements the analysis on location data,and makes users to gather the needed knowledge effectively.