Abstract:The least squares support vector regression estimation method of varying coefficient model（VC-LS-SVR） is a powerful tool recently proposed to analyze the dynamic change characteristics of regression relationship among variables. The new technique of integrating classical non-parametric regression model with machine learning method not only enhances the adaptability and computability of varying coefficient model to the nonlinear relationship of variables, but also inherits the interpretability of classical statistical model. However, this method directly uses the least squares support vector regression（LS-SVR） technology to fit the varying coefficient model, which is lack of robustness. The outliers in the data will distort the estimation of the coefficient function, resulting in misleading interpretation of the regression relationship. In order to get the robust estimation of regression coefficient, this paper proposes two robust estimation methods based on VC-LS-SVR and weighted idea: weighted least squares support vector regression estimation method of varying coefficient model（VC-WLS-SVR） and iterative reweighted least squares support vector regression estimation method of varying coefficient model（VC-IRLS-SVR）. Numerical experimental results show that the two methods are robust and can effectively suppress the influence of outliers. Even when the pollution rate of abnormal data reaches 50% or extreme outliers exist, accurate and stable estimation results can be obtained.

Weighted On-line Support Vector Regression

On-line support vector regression with multiple samples

Weighted Support Vector Machines for Regression and Its Application

Robust Iterative Algorithm of Weighted Least Squares Support Vector Machine and Its Application in Spectral Analysis

A Weighted Support Vector Machine Method and Its Application

A Projection Wavelet Weighted Twin Support Vector Regression and Its Primal Solution

Noise model based ν-support vector regression with its application to short-term wind speed forecasting

Study of on-line weighted least squares support vector machines

Efficient support vector regression with weighted constraints

Wavelet Transform-Based Weighted \(\Nu\)-Twin Support Vector Regression

A New Weighted Support Vector Machine for Regression and Its Parameters Optimization

Application and Study of Support Vector Regression Based on Optimization of Weighted Coefficient

e-Distance Weighted Support Vector Regression

Robust LS-SVR Estimation Algorithm and Numerical Analysis of Varying Coefficient Model

A Weighted One-Class Support Vector Machine

Local least squares support vector regression with application to online modeling for batch processes

On-line Training Algorithm and Its Application Based on Weighted SVR

An effective method for weighted support vector regression based on sample simplification

Weighted proximal support vector machines: Robust classification

Multi-output Support Vector Machine Regression and Its Online Learning.

A Local-Density-Ratio Based Algorithm for Setting Weight in Weighted Least Squares Support Vector Machine