Alternating Projection Algorithm of Linear Regression Method with Convex Constraint in Psychological Dimension Analysis

This paper studies evaluation problem in psychological health. Based on the psychological dimensions and the actual evaluation work, we propose a kind of generalized linear regression model with convex constraint, named as Evaluation Model, in which the dependent variable and regression coefficients all are unknown. We analyze the algorithm of the model with linear regression method. Then the least square estimate of parameters using the alternating projection algorithm is presented. At last, we use this least square iterative method to work out the parameters of the evaluation model. Our work could be a useful guide for the further research of psychological dimension in evaluation work.