Fuzzy Clustering of Incomplete Data by Means of Similarity Measures

Abstract

The current standing in the scope of Data Mining considers clustering to be one of the most useful and widely used tools. Multiple real-world datasets usually contain drops/gaps in the data due to various reasons. The currently known approaches are highly efficient only in those cases when original datasets do not change their volumes during the analysis. However, up-to-date problems mostly deal with sequential online data processing. On the other hand, there is no prior knowledge on which feature vectors contain overlooks. In this manuscript, the challenge of possibilistic and probabilistic online clustering approaches for processing incomplete data is solved through similarity measures of a specific kind which are capable of either loosening outliers’ influence or repressing them.