.png){kind=icon}
.png){kind=icon}
Return
Multiple change-point detection in single index model
.png){kind=icon}
PRE
AI
DOI：10.1080/03610918.2026.2625232.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The single-index model is a classical dimension reduction method with wide-ranging applications. However, real datasets often exhibit heterogeneity, resulting in temporal variations of the linking function. Although previous studies have investigated single change-point detection in single-index models, theoretical results for multiple change-point detection are still lacking. To bridge this gap, we propose two weighted least squares-based multiple change-point estimators, constructed with multivariate and univariate kernels, respectively, and establish their theoretical consistency. Furthermore, a fast iterative estimation procedure is developed to enhance computational efficiency. Simulation studies demonstrate that the proposed estimators achieve performance comparable to state-of-the-art methods. Finally, we illustrate the practical value of the proposed methods by applying them to the analysis of the realized volatility of the NASDAQ index.
Keywords:
Change-point detection
Information criterion
Semiparametric regression
Journal
C
IF:
0.8
Papers:
213
Citations:
4.7K