Applied Research Of Data Mining In The Police Data Analysis System

2013is the first year of age of a large data. Big data has been quietly walked into thepolice and into the new application field. In the process of the public security work on theday-to-day enforcement will accumulate a large amount of data. More and more publicsecurity science and technology department found that police data based on growing scale,has been unable to adapt to the traditional database technology, data analysis and the responsespeed of the cases are getting slower and slower, as a result, the construction of police dataanalysis platform, analyzes all kinds of information resources and video data, mininginformation, from a large database or data warehouse to extract implicit, has potentialapplication value of the information or data, application in the field of public security, toassist in the line of actual combat. The work including the use of huge amounts of data,digging, collision, and the synthesis of a variety of technology, based on data mining, improvepolice informatization, improve work efficiency and better service to the masses.In this article, through the research of data mining techniques and algorithms, thecommissioner’s exit from the data in the data and case data as data sources, in datapreprocessing and multi-dimensional data modeling, dug up some useful information fromdata.In this paper, the main work has the following several aspects:A. Understand the characteristics of the current police data and big data. Public securitybig data is also a kind of "big data", large data volume and fast growth; there are manydifferent kinds of data source are very rich and different structure; Low density scale storage,its value; quite sensitive to time.B. Build police data warehouse. Structure of three layers: the bottom is warehousedatabase server; basic is a relational database system. Data extraction, cleaning, transform,and load and refresh, to update the data warehouse; The middle layer is the OLAP server; Thetop is the front-end client layer, including query and reporting tools, analysis and data miningtools.C. The construction of police multidimensional data model. Basic technology is datacube and OLAP online analysis, and the data of the stratified sample, the classificationmethod for numeric data and non numeric data. The star model was used to constructmultidimensional data model, made up of fact table and dimension table. D. Using the Apriori algorithm and association rule mining outbound data. First of all todo data discrimination, the different property field continues to classification, in order toreduce the workload of data processing, and to facilitate the analysis; set the minimumsupport and minimum confidence of frequent mining, mining frequent item sets. We draw theconclusion base on the analysis of the obtained rules.E. Use the decision tree ID3outbound data mining techniques and algorithms. Thedecision tree USES the root tree node form, tree complexity measure: with the total number ofnodes, leaf number, depth of the tree, and use the number of attributes. ID3algorithm withinformation gain as the decision tree classification attribute standard. By computing theexpected value, the information office, information gain. Choice makes the information GainGain(A) the greatest attributes as the root node of the decision tree, and then the property isdivided into several subsets, a subset of repeated recursive calculation, generation to the nextlevel node, until completion of the split all attributes and concluded the rules.F. Use case data cube generated multidimensional data model, according to the theme ofthe case has multiple dimensions and visualization. Using multi-dimensional statements morecolumns from the dimension of dimension lines and analyzed the cases of data and theanalysis of the corresponding conclusion.