Vague Set Theory In The Association Rules And Cluster Analysis

To deal with fuzzy information, fuzzy sets theory has been founded by Zadeh in 1965. In the next few decades, it continuous development and improvement, and in the field of data mining has been applied, the corresponding fuzzy sets theory based data mining algorithms more practical research and applied projects. However, in subsequent studies, some scholars pointed out that this single value of membership is also inadequate, and fuzzy sets can not be said to handle such information with a fuzzy issue. In order to solve it, vague sets proposed by Gau and Buehrer in 1993, Taiwan scholar, which can be seen as the promotion of Fuzzy Sets. In other words, fuzzy sets is a special case of vague sets, it has a stronger expression of ambiguity and imprecision of data capacity.In this paper, the association rule algorithm based on vague sets was researched. The association rule based on vague sets was defined; a new vague support and confidence based on vague sets were also defined. In the missing data under the Boolean association rules mining, using vague sets to process data, which missing part of the information. It makes more realistic requirements. A new VagueApriori algorithm was presented, which based on vague sets and traditional association rules mining algorithm. The traditional model of Boolean association rule mining problem has expanded the scope to handle missing data under the Boolean association rule mining. It can also handle the traditional problem of mining Boolean association rules. Finally, VagueApriori algorithm and the interaction with the user were implemented using C# programming language. The performance of the algorithm was analyzed by a simple experimental, which validate the efficiency of the algorithm.In addition, the direct clustering method based on fuzzy sets was researched in the paper, and the vague direct clustering method based on vague sets was also researched, which was extended to the vague sets. Some methods to establish vague similar relationship are given, which were implemented using Matlab M-language programming to achieve these calculations process respectively. some methods to construct vague similar matrices are given, the method of draw up web and the max tree method are introduced to vague sets, the concept of graph of vague relations based on the methods are presented, and the direct clustering methods based on vague sets are also given, they are method of vague draw up web and vague max tree method. Then a comparative research was processed between vague direct clustering method and vague equivalent clustering method. Finally, vague equivalent clustering method and the vague direct clustering method are used to calculate respectively by using an example. The experimental results show that the vague direct clustering method is easy, not causing distortion of the original information. It is more effective and more granular classification than vague equivalent clustering method.