Method of Data Clustering Incomplete Fill Based on Constraint Tolerance Set Dissimilarity

In the field of data mining, clustering with loss data is one of the common problems. The traditional approach is based on a probability model to predict the loss data for prediction filling, the limitations of such methods is not suitable for the processing of large amount of data. An improved data clustering incomplete fill algorithm is proposed based on constraint tolerance set dissimilarity, this method does not use the probability hypothesis, and the characteristics of the loss data are used directly. The overall dissimilarity analysis is taken with the set mode, the constraint tolerance difference and the corresponding data object are simplified, the thematic analysis of data set is combined, and the loss data is filled. The baseline data set is used as the experiment sample, the simulation result shows that this method has higher processing speed, the accuracy is improved. It has good application value in data clustering research.