Fuzzy Clustering Mining Technology And Its Application In Applying To College Services

Will-deciding is a critical point in college entrance exam. The students are usually confused without scientific guidance to the will-deciding. Therefore, it is of great significance to provide decision support and intelligence service for will-deciding. Service of recommending academy makes a classification of colleges in accordance with the students' request, and recommends the most suitable colleges to the students. This service reduces the complexity of collecting and analyzing of college information. In this paper, fuzzy clustering techniques in data mining were used to analyze the college data in order to get rational classification result, which get the degree of uncertainty that college belongs to the special class and express an intermediate category. The proposed method made the service of recommending academy more scientific.This paper introduced the relevant basic knowledge of fuzzy cluster analysis, elaborated on the principle and implementation steps of Fuzzy-c means Algorithm. In order to overcome the shortcomings of a great amount of computation while getting the best clustering number using fuzzy c-means algorithm based on validity judge. This paper used subtractive clustering algorithm to determine the maximum number of initial cluster, it reduced the time complexity. The original fuzzy c-means algorithm initialized the cluster centers every time using different cluster number result to the algorithm instable. This paper improved the fuzzy c-means algorithm by using the method of merging cluster centers and overcame the shortcomings of sensitive initialization. In cluster analyzing for academy data, the maximizing deviations method was used to calculate the objective weights of the characteristics and made a characteristics weight by combining the subjective weights set by students firstly. Then the weighted data were clustered by using fuzzy c-means algorithm, it avoided the problem of the homogenizing process for the contribution of different index in cluster process of original fuzzy c-means. The experimental results showed that the improved fuzzy C-means algorithm achieved a better clustering result.In the end of this paper, an academy-recommended model was designed by using the improved fuzzy c-means algorithm, and it was applied into the intelligent service of recommending academy in will-deciding of college entrance exam.