【摘要】 本文考虑两变量随机截距模型在单位正方形设计域上的D-,G-,A-,Ds-和I-最优设计.证明了最优设计在设计区域的顶点处获得,得到了几类不依赖于随机截距项的最优设计模拟结果表明最优设计较随机设计可大幅提高参数估计的精确性.更多还原

【Abstract】 This paper considers optimal designs on[0,1]~2 based on D-,A-,G-,Ds-and I-optimality criteria for bivariate linear regression models with random intercepts.It is shown that the search of optimal designs could be confined at extreme settings of the design region and optimal approximate designs are independent of the variance of random intercepts.Simulation results show that optimal designs can significantly improve the accuracy of parameter estimation than random designs.更多还原