| Data Mining is a cross-disciplinary combinative product, include artificial intelligence, machine learning, database technology and so on, it's a technology to discover unknown information or pattern of potential value from present data. So far, the data mining has been successfully applied to many areas, but it is a beginning to apply it to public security intelligence analysis. What's more, there are many problems to apply the data mining to the public security intelligence analysis. This dissertation introduces self-adaptive clustering algorithm, and improves decision tree algorithm and association algorithm, and then demonstrates its feasibility with experimentation.First, in this kind of cases analysis, traditionally the k-means algorithm was used that need the number of clustering criminal. This method leads to the problem of subjectivity in selection of clustering number and blindness in initial division. On these problems of the algorithm in cases analysis, this dissertation introduces the self-adaptive clustering algorithm to the cases analysis. The results of experiment show that the accuracy of cases analysis is improved. Meanwhile, the introduction of this method will be beneficial to criminal analysis.Then, for the shortage of the ID3 algorithm in objective factors analysis, which ignores the different importance in different crime objective factors, it is brought forward an ID3 algorithm based on weight. It introduces the concept of weight in calculating the information gain; therefore the influence of property selection to classification accuracy is reduced. Introducing the improved ID3 algorithm to objective factors analysis, it is shown that the improved algorithm the validity and effectiveness in the case analysis. At last, for the shortage of the Apriori algorithm in case analysis, which ignores the crucial differences of case property, this paper presents an Apriori algorithm based on weight, which identifies the weight by calculating the information gain. Applying the improved algorithm to case analysis, there is evidence that this algorithm can excavate more valuable rules and the efficiency of algorithm can be approved. |
