A0838
Title: An algorithm for estimating threshold boundary regression models
Authors: ChihHao Chang - National Chengchi University (Taiwan) [presenting]
Abstract: The threshold boundary regression (TBR) model is introduced for the analysis of datasets with binary or continuous responses. By integrating regression models with threshold boundary functions using explanatory variables, the TBR model constructs linear or nonlinear classifiers, partitioning the responses into two groups, with separate regression models fitted to each group. An ordered iterative algorithm called the TBR-WSVM algorithm is proposed to estimate the TBR model. This algorithm combines weighted support vector machine (WSVM) techniques with maximum likelihood and least-squares methods. Through simulation studies and empirical analyses, the performance of the TBR-WSVM algorithm is assessed. The results indicate that the TBR-WSVM algorithm offers robust estimation and prediction capabilities for linear and nonlinear threshold boundary models.
