Title: Estimation of regression parameters using SVM with new methods for meta parameter

Authors: S.S. Desai; D.N. Kashid

Addresses: Department of Statistics, Gopal Krishna Gokhale College, Subhash Road, Kolhapur (MS), 416012, India ' Department of Statistics, Shivaji University, Kolhapur (MS), 416004, India

Abstract: In this article, we propose two methods for selection of meta parameter (C) in support vector regression. The proposed methods are robust because these are based on the robust statistical measures. The performance of the proposed parameter selection methods is evaluated in case of normal and non-normal distributed error variables. It is evaluated in the sense of prediction risk and mean square error of estimates of regression parameters for clean and outlier data.

Keywords: support vector machines; SVM; support vector regression; SVR; meta parameters; prediction risk; mean square error; MSE; parameter estimation; regression parameters.

DOI: 10.1504/IJDMMM.2015.071449

International Journal of Data Mining, Modelling and Management, 2015 Vol.7 No.3, pp.239 - 256

Received: 08 May 2021
Accepted: 12 May 2021

Published online: 28 Aug 2015 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article