A Study On A Grid Clustering Algorithm Based On Adaptive Artificial Immune Evolution

A new grid clustering algorithm based on adaptive artificial immune evolutionary computing is designed. The clustering algorithm is one of data analysis, and could be used in studing of data mining, statistics, machine learning, pattern recognition, biology, spatial database technology, electronic commerce, and failure diagnosis. The clustering algorithm has good action on data clustering and classing. The grid clustering algorithm defines count measure on data space. An objective function is constructed with the count measure which is defined on the space. The value of the objective function could be maximized through adjusting the resolution ratio of dividing the data space. In the computing, the artificial immune algorithm is used to determine the resolution ratio of dividing the data space in the grid clustering algorithm. Thus the resolution ratio of dividing the data space in the grid clustering algorithm is determined by computing, the objectivity of the algorithm is improved. And the applicable scope of the artificial immune algorithm is expanded, the artificial immune algorithm could be used in data mining and solves the key question. The artificial immune algorithm is improved, entropy of information theory is used in the algorithm. The ratio of actual population entropy and the expected value is used as negative feedback, thus the algorithm is adaptive, the performance of the algorithm is enhanced. At the determined resolution ratio, some details of the data are hidden, but key attributes and relatives are more obvious. Thus the data in the same cluster are most similar, and data in different clusters are most different. Cybernetics, information theory and intelligent computation theory are overlapped and seepaged, they are used to solve the actual problum, the applicable scope of cybernetics, information theory and intelligent computation theory is expanded, and cybernetics, information theory and intelligent computation theory are fused. Apriori information of the data from specialists is not necessary, any parameter in the algorithm is not determined by hand, thus the accuracy of this clustering algorithm is improved. The algorithm could cluster data, and its time order of complexity is smaller, and the clustering algorithm is simple, acurate.