纵向数据半参数混合效应的Logistic模型

Logistic model of semiparametric mixed effects for longitudinal data

为了深入探究Logistic模型的统计推断问题,联合半参数混合效应模型和Logistic模型建立了纵向数据半参数混合效应的Logistic模型,进一步求出该模型的极大似然函数、Fisher信息阵,利用QLB(quadratic lower-bound)算法求解参数的极大似然估计.最后进行数值模拟,结果表明在此模型的求解问题上QLB算法是Newton-Raphson算法的合理替代.

In order to semiparametric mixed for longitudinal data, further explore the statistical inference problem of the logistic model, this paper combines the effect model and the logistic model to build a logistic model of semiparametric mixed effects figures out the maximum likelihood function of the model and the Fisher information matrix, and uses the QLB （ Quadratic Lower Bound） algorithm to solve the maximum likelihood estimation of the parame- ters. Finally, the results of the numerical simulation show that the QLB algorithm is a reasonable ahemative to the Newton -Raphson algorithm for such solutions.