Research On Mining Fuzzy Association Rules

Association rules mining is one of the key research content in data mining. There are many different branches of study in association rule mining. The mining of association rules can be divided into two categories according to data attribute types, which are Boolean association rules mining and quantitative association rules mining. Researchers have do more research in Boolean association rule mining and gained a lot of valuable methods. But the study of mining quantitative association rules is still lacking. This paper mainly do some research in mining quantitative association rules. The algorithm has been extended to the fuzzy areas.This paper proposes a BFCFPM algorithm for mining fuzzy association rules, which inherits the recursive thought in FP_Growth algorithm. The main idea of BFCFPM algorithm is as follows. Firstly, mapping the quantitative data sets to fuzzy data sets by FCM clustering technology. Secondly, compress the fuzzy data sets and then store them to IFFPT. At last,recursively mining fuzzy frequent item sets on IFFPT. Loss of information and the rules coursed by discretization is avoided effectively when extracting fuzzy association rules in this algorithm. The algorithm has been tested and has good performance.Another FMFFI algorithm is proposed in this paper. The algorithm is mainly aimed to solute storage problems that BFCFPM algorithm establishes IFFPT when there are more data set records and fewer item number. FCM clustering technology is also be applied to mapping quantitative data sets to fuzzy data sets in this algorithm. When searching for fuzzy item sets,both ways from high-dimensional to low-dimensional and from low-dimensional to high-dimensional are applied. Searching from low-dimensional to high-dimensional adopts Apriori algorithm,FMFFI algorithm also has good performance according to tests.