An Analysis of Fuzzy Weighted Association Rule Mining from Medical Data

Feature selection and classification techniques are the important factors required for the decision making process in Data mining. Recent developments show that fuzzy association rule mining algorithms took the place of conventional classification techniques because of the efficiency of the algorithm. However, fuzzy association rule mining algorithms suffers from the exponential growth of rules produced by fuzzy partitioning of attributes. An attempt is made to propose three different fuzzy weighted association rule mining algorithms. The proposed algorithm incorporates the concept of information gain, gain ratio and correlation value combined with the gain ratio features selection methods. Also a comparative analysis of the proposed algorithms is done with the existing fuzzy association rule mining and it is shown that the classification accuracy value is better for the proposed approaches. The performance has been analyzed with benchmark data collected from the UCI repository.