|
In the real world , many of the boundaries between objective things tend to be vague, the classification of things when they must be accompanied by ambiguity, resulting fuzzy clustering analysis . In the fuzzy clustering analysis is the most typical fuzzy C-means clustering algorithm . First, the data for the interval-valued fuzzy C- means clustering shortcomings , propose an adaptation from the elastic modulus, the interval-valued fuzzy data, improved C- means algorithm , and gives a detailed proof that the improved algorithm using both the traditional fuzzy C-means clustering algorithm, but also consider the size and range of the important properties of the impact level on the clustering results , but also to deal with irregular cluster subset of outliers on the clustering results to avoid damage the clustering results more accurate. Secondly, the database or data warehouse incrementally insert, delete, update the cluster centers proposed method and the class division is automatically determined during the initial cluster centers class division algorithm to solve the insertion , deletion process brings changes in the number of class centers and clustering problem , the algorithm to meet real-time requirements . Finally , comprehensive set theory, the degree of aggregation , separation , update cluster centers algorithm and class division algorithm proposed AIFCM algorithms. AIFCM algorithms in real-time processing incremental problem solving FCM applies only to static data sets defects and AIFCM in the merger process, to previous FCM clustering algorithm can not only find a subset of spherical clusters , can also handle irregular subset clustering structure . AIFCM clustering algorithm can filter the noise , incremental data processing , the user needs to obtain suitable clustering results .
|