Design and Implement Proposed Crime Analysis using Modified Association Rule

Abstract

This research presents a proposal to advance crime analysis that through employee data mining association rules on crime’s data with a proposed strategy consists of three levels, each level present suggestion to suite and consistence crime analysis and predictions. First level will deal with the challenges in mining crime data, where the last often comes from the free text field. While free text fields can give the newspaper columnist, a great story line, converting them into data mining attributes is not always an easy job. The proposal will look at how to arrive at the significant attributes for the data mining systems. That through suggested view organized the crime to three dimensions these are crime attributes, criminal attributes and geo-crime attributes. Second level will use AR (apriori) as a miner technique of crimes, but apriori in case of large dataset is not efficient, also has no security to protect the mined data from unauthorized users. The proposal modify apriori (MAR) to avoid the degradation of performance with crime analysis by reduce unimportant and redundant transactions. Advance MAR with modest suggestion to be secure. Third level, applying the MAR on each dimension separately then according need and on demand of correlate among these dimensions, the correlation done using proposed mixing. The proposal applied on real crime data from a dependable sheriff’s office depended in our previous work (reference 6), then a comparison done between the previous and current work. The results of comparisons show the current work advance previous work by optimizing time and space consumed in mining through apply suggested MAR in current work, where the previous work apply traditional apriori AR. Also the proposed MAR give precision in prediction since it omitting the redundant and ineffective data.