An Epsilon Hierarchical Fuzzy Twin Support Vector Regression

—The research presents -hierarchical fuzzy twin support vector regression (-HFTSVR) based on -fuzzy twin support vector regression (-FTSVR) and -twin support vector regression (-TSVR). -FTSVR is achieved by incorporating trapezoidal fuzzy numbers to -TSVR which takes care of uncertainty existing in forecasting problems. -FTSVR determines a pair of -insensitive proximal functions by solving two related quadratic programming problems. The structural risk minimization principle is implemented by introducing regularization term in primal problems of -FTSVR. This yields dual stable positive definite problems which improves regression performance. -FTSVR is then reformulated as -HFTSVR consisting of a set of hierarchical layers each containing -FTSVR. Experimental results on both synthetic and real datasets reveal that -HFTSVR has remarkable generalization performance with minimum training time.