Designing computer experiments with multiple types of factors: The MaxPro approach
Designing computer experiments with multiple types of factors: The MaxPro approach
复制标题
DOI：
10.1080/00224065.2019.1611351
复制
发表时间：
2020-10
影响因子：
2.5
通讯作者：
V. R. Joseph;Evren Gul;Shan Ba
中科院分区：
文献类型：
--
作者：
V. R. Joseph;Evren Gul;Shan Ba
摘要计算机实验可能不仅涉及连续输入因素，还可能涉及标称因素、离散数值因素和顺序因素。大多数现有的设计计算机实验的文献只关注连续因素。一些工作进一步考虑了名义因素，但也包含离散数值或序数因素的情况几乎被忽略了。在这项工作中，我们提出了一种新的最优设计准则，可以适应所有这些类型的因素。所提出的设计在游程大小和因素数量上是灵活的，并且还可以在全设计空间和所有可能的低维投影中实现良好的空间填充特性。
Abstract Computer experiments may involve not only continuous input factors but also nominal factors, discrete numeric factors, and ordinal factors. Most existing literature in designing computer experiments focuses only on continuous factors. Some works have further considered nominal factors, but the cases that also contain discrete numeric or ordinal factors are almost overlooked. In this work, we propose a new optimal design criterion that can accommodate all these types of factors. The proposed design is flexible in run size and number of factors, and can also achieve good space-filling properties in the full design space and in all possible low-dimensional projections.