Bayesian P-spline estimation of partially linear variable coefficient quantile model
Partial linear variable coefficient model is an important semi-parametric regression model.For the parameter estimation problem of this model,the Bayesian P-spline method is utilized to approximate the unknown smooth function of the non-parametric part,and then the Bayesian quantile regression is implemented using the asymmetric Laplace distribution to derive the posterior distributions of all the unknown parameters in order to obtain the estimates of the parameters,the parameter estimates were obtained by Gibbs and Metropolis-Hastings algorithm.Meanwhile,the estimation effect of Bayesian P-spline method is compared and analyzed with B-spline method through numerical simulation,and the results show that the Bayesian P-spline method has better estimation effect at different quartiles under the mean square error and standard deviation criterion.
partial linear variable coefficient modelBayesian P-splineB-splineGibbs samplingMean square error
万方数据
维普
