Title: Hermite polynomial estimation of index models after a nonparametric transformation
Authors: Yundong Tu - Peking University (China) [presenting]
Abstract: The purpose is to estimate transformation models with the index structure by using Hermite Polynomials. The estimation is performed using least squares, which is computational efficient for practical implementation. Asymptotic properties of the estimators of the index parameter, the unknown transformation function, the unknown link function are studied. Simulation studies show that the estimators enjoy nice finite sample properties and outperform the competitors available in the literature. The procedure is illustrated with a real data example.