Minimum spanning tree clustering algorithm and system based on density core

Abstract

The invention provides a minimum spanning tree clustering algorithm and system based on a density core. The algorithm comprises the following steps: constructing a KD tree; obtaining reverse neighborinformation and natural characteristic values of the data points by adopting a natural neighbor method, and counting the number of reverse neighbors of each data point; taking the data points of whichthe reverse neighbor number is not less than the natural characteristic value as core points, and forming a density core point set by the core points; establishing a minimum spanning tree according to the density core set to obtain a set of weights of each edge in the minimum spanning tree; calculating a trimming threshold value according to the set of the weights of the edges in the minimum spanning tree, and cutting off the edges connected with different clusters in the minimum spanning tree according to the trimming threshold value to obtain a minimum spanning sub-tree of each sub-cluster;generating a sub-tree clustering density core according to the obtained sub-cluster minimum; and distributing the non-density core points to the cluster of the density core closest to the non-densitycore points to complete clustering. According to the algorithm, the approximate shape and structure of the cluster can be well reserved, so that the algorithm can adapt to a data set with a complex shape.

Description

technical field [0001] The invention relates to the field of computers, in particular to a minimum spanning tree clustering algorithm and system based on a density kernel. Background technique [0002] In recent years, data information has shown an explosive growth trend, forming huge and complex massive data. These valuable data cannot be analyzed and understood by humans without the help of automatic analysis methods. In order to extract effective information in a timely manner from huge and complex data, effective data analysis tools are needed, and data mining technology has emerged along with the trend. [0003] Data mining can be divided into the following three steps: preprocessing, data analysis and knowledge evaluation. After obtaining the original data from the data source, use data preprocessing methods such as data cleaning, data conversion, feature selection, and feature extraction for preprocessing, and then select data analysis algorithms to analyze the proc...

AI Technical Summary

Problems solved by technology

The density peak clustering algorithm can effectively cluster data sets, but it needs to manually select the cluster center, cannot handle data sets with complex shapes, and has disadvantages such as high time complexity

Images