Kernel-based Adaptive Fuzzy C-means Clustering Algorithm Based On Fruit Fly Algorithm And Association Rule Mining

Clustering analysis and association rule mining has been focused on domestic and foreign experts and scholars as important research direction of data mining technology. This paper focus on the fuzzy C mean clustering algorithm (Fuzzy C-means, FCM) and the apriori algerithm. the main work is as follows:An adaptive KFCM clustering method based on fruit fly optimization algorithm is proposedFCM algorithm is more suitable for practical application due to fuzzy membership.N(?)days. it has become one of the widely used clustering algorithms. However, the traditional FCM clustering algorithm has some limitations in practical applications, for example, the(?) is sensitive to the initial value, easy to fall into the local minimum and vulnerable to(?) distribution and fuzzy parameters. According to the existing problems, a method of(?)-based adaptive fuzzy c-means clustering based on fruit fly algorithms (FOAKFCM) is proposed in this paper. In this algorithm,firstly,the Gauss kernel function is used to optimize the objective function of the FCM algorithm, which is used to transform the discrete attribute data from the low dimensional feature space to the high dimensional feature space, and to expand the difference between features, then the iterative process of the fruit fly optimization algorithm is used to replace the iterative process of kernel-based FCM algorithm (KFCM)algorithm. Secondly, the clustering validity evaluation index MIA is introduced to select the fuzzy parameters of KFCM. Experimental results show that the proposed algorithm improves the clustering accuracy of FCM algorithm and the clustering effect is better.2. The improved apriori algorithm based on location storage is proposedApriori algorithm has been applied in many fields as the classical algorithm of association rules, and it has been used to deal with the combinatorial explosion problem caused by frequent itemsets successfully in the early stage. However,this algorithm leads to a large amount of time and space because it generates large number of candidate itemsets and the multiple scan of the database. Aiming at the limitation of the algorithm, an Apriori algorithm based on location memory (L-Apriori) is proposed in this paper. First, the new algorithm scan the database to construct matrix, and then in order to reduce the run time and memory space, we generation of candidate itemsets by transform position which nonzero element coordinates in a matrix, and dynamic pruning. The test results show that the proposed algorithm can reduce the time and space of the algorithm effectively, and improve the performance of the Apriori algorithm.3. The application of FOAKFCM algorithm and L-Apriori algorithmThe two algorithms are combined and applied to the mining of fuzzy association rules. Firstly,the adaptive KFCM algorithm (FOAKFCM) based on fruit fly optimization algorithm is used to preprocess the numerical data, from here we obtain fuzzy partition and data membership, after that, the improved Apriori algorithm (L-Apriori algorithm) is used to mining the association rules. The experimental results show that the mining association rules are strongly correlated,the effectiveness and feasibility of the plan that two algorithms in fuzzy association rule mining are verified.