A0703
Title: Trigonometry-transformation based correlation coefficient with an application to sufficient variable selection
Authors: Pei Wang - Bowling Green State University (United States) [presenting]
Abstract: The technique of variable selection has gained widespread popularity for reducing data size, particularly in the context of large p small n datasets. A novel criterion is introduced based on the correlation coefficient derived from trigonometry transformations. This innovative criterion serves as a metric for assessing the relationship between the response and each predictor. When integrated into a two-step selection procedure, it becomes a valuable tool for variable selection. Notably, this approach is model-free, providing robustness against model misspecification. The asymptotic and sure selection properties are established, and the effectiveness of the proposed method is demonstrated through extensive numerical studies and real data analysis.
