A Constrained Bicluster Mining and Missing Value Prediction Method Based on Preserving Submatrix

Abstract

The invention discloses a constrained bi-clustering mining and missing value prediction method based on an order-preserving sub-matrix, comprising the following steps: S1, receiving a request and mining the bi-clustering mode of the data matrix; S2, calculating the overlapping degree of rows and columns and coverage; S3, judging whether the coverage of rows and columns converges, if not, proceed to step S4, otherwise proceed to step S9; S4, randomly disrupt the order of rows and columns in the data matrix, and sort the set of candidate columns; S5, judge the set of candidate columns Whether it is empty; S6. Judging whether there are qualified columns in the candidate column set; S7. Expanding the column set and row set of the double-clustering mode according to the newly selected qualified columns; S8. Judging whether the current double-clustering mode conforms to the sparse order Requirements of the sub-matrix model; S9, calculating the missing value of each bi-clustering pattern; S10, returning the final bi-clustering set and missing value. Therefore, the present invention effectively improves the quality of the biclustering model and the accuracy of missing value prediction.

Description

Technical field[0001]The present invention relates to the field of dual clustering mode excavation, and more particularly to a constrained dual clustering mining and lack value prediction method based on a sequestrative sub-matrix.Background technique[0002]The general clustering is based on all attributes of the data, and the data cluster is called conventional clusters. Traditional clusters can only look for global information and cannot find local information, while a large amount of interesting information is hidden in these local information. In order to better search local information in the data matrix, people present a double clustering concept. The so-called dual cluster, also called sub-matrix mode, is a set of row sets with local similarities in the data matrix, a sub-matrix of a set of column subset, which means that this group is set on this set of columns. Performance has some local similarity. The dual cluster excavation is to capture interesting local similarity in th...

AI Technical Summary

Problems solved by technology

However, these methods do not scale well when the data matrix becomes very large

Furthermore, no overlapping or coverage constraints are taken into account during pattern mining of order-preserving submatrixes, nor does it address the issue of sparse matrices

Images