On symmetrical equidistant minimax designs in statistical experiments with binary data
On symmetrical equidistant minimax designs in statistical experiments with binary data
复制标题
DOI：
10.1081/sac-120037247
复制
发表时间：
2004-01-01
影响因子：
0.9
通讯作者：
Seidel, W
中科院分区：
文献类型：
--
作者：
Begun, A;Seidel, W
在非线性回归模型中，实验设计的标准最优函数通常依赖于未知参数。局部最优设计对于未知参数的错误规范不具有鲁棒性。它们可以用极大极小法进行鲁棒化。通过对目标函数的全面分析，可以减少内部优化和外部优化的搜索空间。基于对考虑的目标函数的理论和实证研究，提出了一种高效可靠的双参数logistic剂量-响应模型中平衡等距C和d最优极大极小设计的计算算法。
In nonlinear regression models, standard optimality functions for experimental designs usually depend on the unknown parameters. Locally optimal designs are not robust with respect to misspecifications of unknown parameters. They can be robustified by using a minimax approach. A comprehensive analysis of the goal function makes it possible to reduce the search space for the inner as well as the outer optimization. We present an efficient and reliable algorithm for calculating balanced equidistant C- and D-optimal minimax designs in a two-parameter logistic dose-response model, which is based on theoretical and empirical investigations of the considered goal functions.