Probit回归模型参数的约束EM算法

The restricted EM algorithm on the parameters for Probit regression model

针对Probit模型提出了n+1-参数的约束EM算法,其中模型的解释变量含有缺失数据,回归参数含有不等式约束条件。EM算法的第M步采用牛顿迭代法,并给出了算法的收敛性证明。

We proposed the restricted EM algorithm for the Probit model with n＋l-parameters, The explanatory variables in the model which contain the missing data, and regression parameters containing inequality constraints. The Newton Method is used in the M step for the EM algorithm, and the convergence of the algorithm is proved.