Estimation on Generalized Semi-varying Coefficient Models

The procedures for estimations of the parametric component and the nonparametric component on generalized semi-varying coefficient models are developed as follows: The coefficient functions in the parametric component are estimated via local quasi-likelihood method and average method. After replacing the coefficients in the parametric component by the obtained estimators, quasi likelihood method and backfitting technical are employed to produce the estimators of the function coefficients in nonparametric part. For this procedure, the asymptotic normalities of the estimations on parametric component and nonparametric component are investigated. Two simulated examples show that the procedure is effective.