An attribute weighted fuzzy c-means algorithm for incomplete data sets

In many areas, including natural science and engineering technology, data sets are often plagued by the unavoidable problem of data incompleteness. Therefore, the problem of clustering incomplete data sets has become one of the research focuses in the field of pattern recognition, however, the existing algorithms that cluster incomplete data generally assume that each attribute plays a uniform contribution. To overcome this disadvantage, an attribute weighted fuzzy c-means algorithm for incomplete data clustering is proposed in this paper, in which the important degree of each attribute are viewed as additional variables and optimized during clustering. And attribute weights, missing attribute values and clustering results can be obtained simultaneously. The experimental results show that the proposed algorithm can emphasize the important attributes in clustering, and better clustering results can be obtained.