Regularized directional regression models
Conference
65th ISI World Statistics Congress 2025
Format: IPS Abstract - WSC 2025
Keywords: circular, directional data, environmental, regression, regularization
Session: IPS 779 - Advances in Directional Statistics
Monday 6 October 2 p.m. - 3:40 p.m. (Europe/Amsterdam)
Abstract
Circular regression models are widely used in several areas, including meteorology, biology, and geology. The issue of regularization in circular regression models remains an intriguing subject. Least-squares estimation has been the predominant approach for determining regression parameters in circular data. However, this approach does not produce accurate results in the presence of multicollinearity. This study focuses on response prediction and regularized circular regression models. Alternatives to the least-square estimators for the circular-linear regression model are suggested. A comparative study is conducted on simulated data to assess the least-square estimators and proposed methods. Model validation is performed to evaluate the model’s adequacy. The proposed methods are applied within the environmental domain.