Extended Twin Support Vector Regression via Particle Swarm Optimization for the Interval Data

• Main Authors: CHEN, CHIA-WEI, 陳嘉緯
• Other Authors: HSU, HSIANG-LING
• Format: Others
• Language: en_US
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/4n36m4

碩士 === 國立高雄大學 === 統計學研究所 === 106 === For the past decade, generalized support vector regression methods have been applied to model the interval-valued data. Based on the idea of support vector regression (SVR) and twin support vector regression (TSVR), Peng et al. (2015) proposed a novel interval twin support vector regression (ITSVR), which employs two nonparallel functions to identify the upper and lower sides of the interval output data. For analyzing data more flexible, we construct a so-called extended twin support vector regression (ETSVR) method, which concept utilizes two related TSVR functions to model the upper and lower ends of the interval data. In the meantime, we adopt the Particle Swarm Optimization (PSO) algorithm to reduce the searching burden for the tuning parameters of our algorithms. The analyzed results of the artificial and real data in this study show that the constructed interval models by ETSVR gives better performances than those based on ITSVR.