An optimal scheme selection method based on fuzzy clustering iteration and projection pursuit

Abstract

The invention discloses an optimal scheme selection method based on fuzzy clustering iteration and projection pursuit in the technical field of data clustering mining. The specific steps of the methodare as follows: S1, moderately evaluating each individual in a sample set to find the optimal object; S2: parallel evolution of two spaces; 3, after that main body group space and the belief space are mutually fused, if the evolution index reach an integral multiple of the set value, the receiving operation and the influence operation are carried out; 4, stopping that operation when the operationtermination condition is reached, otherwise, the iteration numb is increased by 1, and the operation is continued; 5, selecting that optimal scheme output according to the short distance between sample parts, The invention can avoid the selection problem of the density window width, intelligently sets all parameters in the operation process, ensures the clustering effect through the projection pursuit clustering and the fuzzy clustering iteration, optimizes the clustering center and the projection direction, improves the optimization speed of the data key vector and the selection time of theoptimal scheme.

Description

technical field [0001] The invention discloses an optimal scheme selection method based on fuzzy clustering iteration and projection pursuit, and specifically belongs to the technical field of data clustering mining. Background technique [0002] Clustering is an essential subfield in data mining. The purpose of the clustering algorithm is to gather each type of similar samples in a sample set into clusters, which can be described as a connected area in a multi-dimensional space containing a relatively high density point set. Regions with relatively low point sets are separated from other regions. For clustering algorithms, it is required that the similarity of samples within the same class be as large as possible, and the similarity between different categories be as small as possible. Generally, the two parameters of intra-class density and inter-class distance described according to the distance between samples are used to cluster The effect is judged, that is, the larg...

AI Technical Summary

Problems solved by technology

[0003] However, the traditional projection pursuit clustering uses the standard deviation of projection values ​​to represent the inter-class density, and the distribution of projected points in each window to represent the intra-class density. The density window width is the only parameter that needs to be set, and its value Whether it is reasonable or not is directly related to the validity of the clustering results. At present, the calculation methods used to determine the density window width lack theoretical proof and cannot verify the scientific nature of the clustering results; and the number of samples to be evaluated in the fuzzy clustering iteration is large and the index When the dimension is high, its computational complexity is very high, and it is highly dependent on the dispersion of samples. It is greatly affected by the preset cluster center, and it is easy to fall into local convergence.

Images