| Mining association rules describe the interesting connection between data items and data entry and tap out the rules to guide the user to decide. For the traditional association rules mining algorithm research focuses on frequent item sets and rules for optimizing excavated two pruning, but mining association rules out is a common problem:the huge rule set, multiple redundant rules, slow system speed low efficiency, the rules are difficult to understand and implement decisions users. Therefore, when faced with such a huge set of rules, we usually work to be done is to analyze the measure of each rule to determine which is the reference value for the user, so that the work is undoubtedly inefficient and boring. Current data mining research hot spot is how to mine the rule efficient and useful. Based on this issue, in the study of this paper, we introduce the concept of combination patterns between attributes by mutual combination between the rules, and the rules of the combined evaluation, the final extract useful and user decision support knowledge.This paper describes the data mining association rules and basic concepts, and typical algorithms of association rules for the relevant description and analysis. In measures of association rules, first elaborated the traditional "support-confidence" system to analyze the pros and cons, and then a huge problem for the traditional set of rules generated by the algorithm, we introduce the concept of the degree of interest, by degree of interest to combine rules cluster generation rules and regulations, and the corresponding principle and algorithm steps. Finally, the theoretical results of this study applied to real data sets supermarkets, supermarket data by extracting, cleaning and finishing, and the use of specialized data mining tools AlphaMiner mining results demonstrate; through the analysis of the results obtained can be used decisions on knowledge, thus demonstrating the scientific and effectiveness of the method. |
PDF Full Text Request