A block padding approach in multidimensional dependency missing data. (April 2023)

• Record Type: Journal Article
• Authors: Xu, Huiyan; Chen, Yanli
• Is Part Of: Engineering applications of artificial intelligence. Volume 120(2023)
• Publication Date: 2023-04
• Subjects: High dimensional correlation data -- Missing data -- Block imputation -- Optimization -- Rough set; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2023.105929
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Without high-quality data, there will be no high-quality data mining results. Lack of data value is one of the problems often encountered in data analysis. The filling method of missing data under high-dimensional correlation is researched in the paper. Based on rough set theory of artificial intelligence, a new method blocks is proposed for filling missing data, including ideas, definitions, theorems and algorithms. The relationship between variables is used to fill in missing data, reduce the impact of irrelevant data on missing data, and improve data filling accuracy. In order to segment missing data with unknown block state, blocks are clustered based on k-means. The block filling algorithm can be used for gene sequencing data. Simulation experiments and actual dataset experiments show that in the block filling method, the relevant information in the filled data can be effectively used, and the data filling accuracy is improved.