Intuitionistic fuzzy C-regression by using least squares support vector regression

•We developed a novel clustering algorithm for solving actual clustering problems.•The novel clustering algorithm improves conventional fuzzy c-regression model.•Empirical results indicate that the proposed clustering algorithm has superior performance. This paper proposes a novel intuitionistic fuz...

Full description

Saved in:
Bibliographic Details
Published in:Expert systems with applications Vol. 64; pp. 296 - 304
Main Authors: Lin, Kuo-Ping, Chang, Hao-Feng, Chen, Tung-Lian, Lu, Yu-Ming, Wang, Ching-Hsin
Format: Journal Article
Language:English
Published: Elsevier Ltd 01.12.2016
Subjects:
Fuzzy c-regression model
Intuitionistic fuzzy sets
Least squares support vector regression
Sammon mapping
Intuitionistic fuzzy sets
Sammon mapping
Least squares support vector regression
Fuzzy c-regression model
ISSN:0957-4174, 1873-6793
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:•We developed a novel clustering algorithm for solving actual clustering problems.•The novel clustering algorithm improves conventional fuzzy c-regression model.•Empirical results indicate that the proposed clustering algorithm has superior performance. This paper proposes a novel intuitionistic fuzzy c-least squares support vector regression (IFC-LSSVR) with a Sammon mapping clustering algorithm. Sammon mapping effectively reduces the complexity of raw data, while intuitionistic fuzzy sets (IFSs) can effectively tune the membership of data points, and LSSVR improves the conventional fuzzy c-regression model. The proposed clustering algorithm combines the advantages of IFSs, LSSVR and Sammon mapping for solving actual clustering problems. Moreover, IFC-LSSVR with Sammon mapping adopts particle swarm optimization to obtain optimal parameters. Experiments conducted on a web-based adaptive learning environment and a dataset of wheat varieties demonstrate that the proposed algorithm is more efficient than conventional algorithms, such as the k-means (KM) and fuzzy c-means (FCM) clustering algorithms, in standard measurement indexes. This study thus demonstrates that the proposed model is a credible fuzzy clustering algorithm. The novel method contributes not only to the theoretical aspects of fuzzy clustering, but is also widely applicable in data mining, image systems, rule-based expert systems and prediction problems.
ISSN:0957-4174
1873-6793
DOI:10.1016/j.eswa.2016.07.040