The Research Of Fuzzy Clustering Algorithm Of Data Mining Based On Shuffled Frog Leaping Algorithm

In the last decade, with the rapid development and wide application of computertechnology, communications technology and network technology, enterprises arefacing a growing number of business data. On the one hand, these rich data provideenterprises with information for decision-making can bring the business profits. Onthe other hand, the scientific research of the production process based on largeamounts of data. we need to analyze and deal with the data so that we can achievethe process of identification, fault diagnosis and control of decision-making of theproduction process. Therefore, data mining as one of the main techniques to locatethe useful information in large, complex data causes widespread attention inacademia and industry.Shuffled frog leaping algorithm simulates the intelligent search behavior offrog swarms.SFLA has been paid more and more attention by many scholars becauseof few control parameters, easy to implement and simple programming.FuzzyC-means clustering (FCM) and kernel Fuzzy C-means clustering (KFCM)algorithm have been applied to pattern recognition, image processing and manyother fields, but there are still some flaws.Because of the problems such as thesensitivity to initial value and and its ready occurrence of local minimum,A fuzzyC-means clustering based on shuffled frog leaping algorithm (SFLA-FCM) ispresented in this paper.Although KFCM to a certain extent overcomes the data shapedependence, it still exists some drawbacks,like FCM algorithm,such as thesensitivity to initial value and the shortcomings of the local minima. Therefore, ankernel fuzzy C-Means based on shuffled frog leaping algorithm (SFLA-KFCM)isproposed.SFLA-KFCM can obtain ideal clustering results for a small number of datasets.However, its effect was not satisfactory for the data with larger clusters numberand higher dimensions.So adaptive inertia weight is used to update the strategy ofSFLA.Then the obtained optimal solution by adaptive inertia weight shuffled frogleaping algorithm was taken as initial clustering centers of KFCM algorithm tooptimize initial clustering centers,so as to get the global optimum and overcome theshortcoming of the KFCM algorithm.SFLA-FCM can obtain ideal clustering results for a small number and the lowdimensionality of data sets. However, its effect was not obvious for the data withlarger clusters number and higher dimensions.Two improved shuffled frog leapingalgorithm are presented.In this algorithm.The first,the linear decreasing inertia weight is introduced to correct the poor frog update strategy.Then the frog withbetter fitness value is selected to substitute the poor one,and make very frog tomutate with different probability.The second, frog population is initialized with Tentchaotic sequence to enhance the diversity of groups and improve the quality ofinitial solution.Then selected the different mutation probability according to thefitness of each frog population fitness variance.The optimal solution obtained withtwo improved SFLA with strong global searching ability was taken as initialclustering centers of FCM algorithm to optimize initial clustering centers,so as to getthe global optimum and overcome the shortcoming of the FCM algorithm.