Inference for change-plane regression
We establish the asymptotic behavior of change-plane estimators. Two types of estimators of the change-plane parameters are proposed to address the non-uniqueness of the argmins of the objective function. The n rate of convergence and a full characterization of the limiting distributions of the change-plane estimators are shown through theoretical exposition and numerical simulations. A parametric bootstrap procedure is introduced for inference. We use the model to analyze the ACTG175 AIDS study data for precision medicine.
READ FULL TEXT