Imputation of incomplete data using adaptive ellipsoids with linear regression

Article type: Research Article

Authors: Yao, Leehter^* | Weng, Kuei-Sung

Affiliations: Department of Electrical Engineering, National Taipei University of Technology, Taiwan, R. O. C.

Correspondence: [*] Correspondence to: Leehter Yao, Department of Electrical Engineering, National Taipei University of Technology, 1, Sec. 3, Chung-Hsiao E. Rd., Taipei 10608, Taiwan, R. O. C. Tel.: +886 910 182 525; Fax: +886 2 2751 8845; ltyao@ntut.edu.tw

Abstract: An efficient scheme for imputing the features missing of incomplete data is proposed in this paper. The missing features are imputed based on a group of nearest complete data in the space of residual features of the incomplete data to be recovered. In order to find the complete data points in the space of residual features, an algorithm called the evolutionary Gustafson-Kessel algorithm (EGKA) is proposed that learns the ellipsoid to adaptively cluster the complete data points with the recovered incomplete data points. A linear regression model is utilized to impute the missing features based on the complete data clustered by the ellipsoid learned by the EGKA.

Keywords: Incomplete data, fuzzy clustering, particle swarm optimization, Gustafson-Kessel algorithm, linear regression

DOI: 10.3233/IFS-151592

Journal: Journal of Intelligent & Fuzzy Systems, vol. 29, no. 1, pp. 253-265, 2015