Design And Implementation Of Crime Prediction Model Based On Data Mining Technology

With the rapid development of economy,technology and population, China's social security situation has changed significantly since the 21st century. Crime becomes more and more dynamic and intelligent which gives new challenge to police work. Therefore, The Ministry of Public Security initiates the strategy of "Intelligence-Led Policing" and promote the police information construction and application based on "golden shield project". After ten years'hard work, the police information construction has made great progress. Not only large amounts of people, belongings, criminal cases and behavior data have been collected, but also data applications have been explored.In this case, the thesis studies intensively police information data application in crime prediction based on data mining technology. It analyzes the internal association between objects and crime behaviors, and establishes crime prediction model which can predict high-risk people. The thesis concludes:First, the thesis reviewed the current research status of the study and application of data mining technology in crime prediction in China and foreign countries. It also expounded the concept, task, function, mining procedure and main algorithm of data mining technology theoretically and interpreted crime prediction principle from the view of behavioral and social learning theory.Secondly, it analyzed the status of police data resources in detail from the aspects of data structure, data quality and data relationship. It also combined results with the needs of later data mining, and accomplished data preprocessing. In addition, the thesis solved data deletion, data noise and data inconsistency which existed objectively in police information by using data cleaning, data integration, data transformation and data reduction. It laid the foundation for mining work.Thirdly, to meet the actual needs of police work, the thesis divided police information into two parts, static police information and dynamic police information. It first analyzed static police information by using association analysis mining and studied the concepts and principles of association analysis intensively. It adopted Apriori Algorithm neatly to study four key questions in police information analysis. In this way, it got 16 rules which the Public Security Organs was interested in, and established crime prediction model. Then, the thesis analyzed dynamic police information by using naive Bayes classification technology whose concepts and principles was studied intensively. It got prior probability and Conditional probability table by analyzing training dataset, and established naive Bayes classification which was proved correct by raining dataset and functional by practice.Fourthly, the thesis used two models to design crime prediction system which mainly satisfied the requirements of police work. According to the cue from the system, the police can take the corresponding measures to prediction people. It may predict high-risk people, and give precision strike to criminals.In a word, the thesis obtained prediction model and designed crime prediction system by studying crime prediction deeply based on data mining technology. The accuracy of model and reliability of system have been proved feasible by reserved data and work experience. It can improve police work efficiency and strengthen the polices' capability of controlling public security.