Clustering method of incremental algorithm, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a clustering method of an incremental algorithm, electronic equipment and a storage medium, and is used for a combined clustering model of divided clustering and hierarchical clustering, the model can quickly obtain a clustering result with a large number of clustering numbers through divided clustering, and hierarchical clustering of a target clustering number is carried out on the basis, so that the clustering efficiency is improved. According to the combined clustering model, repeated clustering can be avoided when the clustering number is changed, and time consumption is reduced. The method provided by the embodiment of the invention comprises the following steps: during first clustering, performing division clustering on a first data set by using a MiniBatchK-means algorithm to obtain a clustering model; inputting the second data set into the clustering model, adjusting a clustering center point, carrying out clustering identification, and outputting a preliminary clustering result of a clustering number k = m; inputting the preliminary clustering result into a hierarchical clustering model, and outputting a clustering result nlt of a target clustering number k = n; m, n and m are positive integers.

Description

technical field [0001] The present application relates to the field of computer algorithms, in particular to an incremental algorithm clustering method, electronic equipment and storage media. Background technique [0002] Clustering is to divide data points into groups, and at the same time make the similarity between data points in the group as large as possible, and the similarity between data points in different groups as small as possible. Clustering algorithms include partitioning clustering, hierarchical clustering, density clustering, etc. The time complexity of partitioned clustering is low, but the number of clusters needs to be given in advance, and each time the number of clusters is changed, clustering needs to be re-clustered. Hierarchical clustering and density clustering have high time complexity and are not suitable for large data clustering. [0003] When using the classic K-means clustering algorithm to cluster data with a large amount of data and high d...

AI Technical Summary

Problems solved by technology

The time complexity of partitioned clustering is low, but the number of clusters needs to be given in advance, and each time the number of clusters is changed, clustering needs to be re-clustered

Hierarchical clustering and density clustering have high time complexity and are not suitable for large data clustering

[0003] When using the classic K-means clustering algorithm to cluster data with a large amount of data and high dimensions, due to the generation of too many intermediate variables, such as the distance between data points, etc., it will cause memory overflow problems; And the completed clustering can only divide the newly added data, but cannot update the cluster; secondly, every time the number of clusters is changed, clustering needs to be re-clustered, which is inefficient

Images