1: \begin{abstract}

2: 

3: In this paper, we study the model selection and structure specification

4: for the generalised semi-varying coefficient models (GSVCMs), where the

5: number of potential covariates is allowed to be larger than the sample

6: size. We first propose a penalised likelihood method with the LASSO

7: penalty function to obtain the preliminary estimates of the functional

8: coefficients. Then, using the quadratic approximation for the local

9: log-likelihood function and the adaptive group LASSO penalty (or the

10: local linear approximation of the group SCAD penalty) with the help of

11: the preliminary estimation of the functional coefficients, we introduce

12: a novel penalised weighted least squares procedure to select the

13: significant covariates and identify the constant coefficients among the

14: coefficients of the selected covariates, which could thus specify the

15: semiparametric modelling structure. The developed model selection and

16: structure specification approach not only inherits many nice

17: statistical properties from the local maximum likelihood estimation and

18: nonconcave penalised likelihood method, but also computationally

19: attractive thanks to the computational algorithm that is proposed to

20: implement our method. Under some mild conditions, we establish the

21: asymptotic properties for the proposed model selection and estimation

22: procedure such as the sparsity and oracle property. We also conduct

23: simulation studies to examine the finite sample performance of the

24: proposed method, and finally apply the method to analyse a real data

25: set, which leads to some interesting findings.

26: \end{abstract}