A visualization method of fuzzy clustering results based on radviz

Abstract

The present invention provides a kind of fuzzy clustering result visualization method based on Radviz, and its steps are: 1) carry out data preprocessing to the result of fuzzy clustering algorithm; 2) design reasonable clustering cluster dimension anchor point layout for RadViz circle; 3) Project the data samples into Radviz in dot mode or pie chart mode; 4) Expand the dimension anchor point into a dimension ring to realize the visual coding of the Radviz ring; 5) Integrate the membership degree distribution information into the main Radviz 6) Extract the coexistence relationship between clusters, and use the chord to map the coexistence relationship. Based on Radviz, the present invention displays the fuzzy membership matrix obtained by the fuzzy clustering algorithm, which not only provides users with as much fuzzy clustering information as possible, but also allows researchers to freely explore information such as data attributes, membership matrix, and cluster clusters , so that users can make more rapid, intuitive and accurate decisions.

Description

technical field [0001] The invention belongs to the technical field of computer information processing, and relates to a visualization method of fuzzy clustering results based on Radviz. Background technique [0002] Fuzzy clustering is an overlapping clustering method that allows data objects to belong to more than one cluster. In fuzzy clustering, data points correspond to a degree of membership on each cluster, which reflects the degree of uncertainty that the data point belongs to the category, and the result of fuzzy clustering is that the data point belongs to each cluster membership matrix. Since it is difficult to accurately determine the boundaries of clusters in the real world, the partial division of fuzzy clustering is more reasonable and can more objectively reflect the real world. However, when the clustering algorithm obtains many clusters and the data contains a large number of nodes, the membership matrix is ​​a high-dimensional data with a complex structu...

AI Technical Summary

Problems solved by technology

There are some inherent defects in the process of transforming fuzzy membership into hard judgment, such as too much information loss, relatively large errors may occur, and even the opposite result is obtained.

[0004] On the other hand, the membership degree matrix stores the membership degree of the data points divided into each cluster, which can help us intuitively judge which cluster the data point should be divided into relatively speaking, but it does not allow us to obtain more More meaningful information, such as the relationship between multiple clusters, the size of each cluster, etc.

[0005] These problems will make it difficult for users to fully and accurately understand the results of fuzzy clustering, and to some extent weaken the ease of use of fuzzy clustering analysis in practical applications.

Images