Open AccessSETAR Model Research Based on Support Vector Regression Parameter Estimation Method and Empirical Analysis
Author(s) -
Yun Li,
Xiuxia Yin,
Xueting Wen
Publication year - 2022
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2216/1/012093
Subject(s) - setar , support vector machine , statistics , mathematics , econometrics , data mining , computer science , machine learning , time series , star model , autoregressive integrated moving average
In this paper, we use support vector regression (SVR) to estimate the parameters of the SETAR model for predicting the log-return of stock price data. SETAR model is usually estimated using maximum likelihood (ML) and minimize conditional sum-of-squares (CSS), assuming that the data are normal distribution. SVR is a powerful tool for regression estimation. Therefore, this paper proposes to use SVR instead of maximum likelihood and conditional least squares to estimate the parameters of the SETAR model. Empirical analysis on the two stock price data of China Unicom and ICBC, the results show that on the data of China Unicom, the prediction error of the Gaussian kernel SVR-SETAR model is 8% lower than that of the ML-SETAR model and the CSS-SETAR model. And on ICBC data, the prediction error of the Gaussian kernel SVR-SETAR model is 26.32% and 20% lower than that of the ML-SETAR model and CSS-SETAR model, respectively. This suggests that SVR can be used to estimate the parameters of the SESAT model and the SVR-SETAR model has higher predictive power.