Classification for Electric Power Companies Based on Fuzzy Clustering Algorithm

Fuzzy clustering is an important approach in data mining. It has been applied broadly in many aspects and receiving great attention from enterprisers and scholars. This paper makes use of MATLAB language to produce a fuzzy clustering algorithm for classifying electric power companies according to some general financial indexes at the end of 2006, such as ratio of operating income, ratio of stockholder´s equity and current ratio. By constructing and normalizing initial partition matrix, getting fuzzy similar matrix with Minkowski metric and gaining the transitive closure, the dynamic fuzzy clustering analysis for electric power companies is shown clearly that different clustered result change gradually with the threshold lambda reducing. It has a great value on many things, such as contrasting electric power companies´ financial condition in order to grasp the chance of investment.