OptimaRegion is a package for the computation of confidence regions on the location of the optima of response surface models. Both parametric (quadratic polynomial) and nonparametric (thin plate spline) models are supported. The confidence regions obtained do not rely on any distributional assumption, such as Normality of the response, and are obtained by bootstrapping. The resulting regions are both valid and unbiased, and have a size that rapidly decreases as the sample size increases. Regions are obtained both numerically (as a set of points) and graphically, as the convex hull of the points. Functionality for the computation of a bootstrap confidence interval on the distance between the optima of two different response surfaces is included.
Del Castillo, E., Hunt, J., Rapkin, J., and Zarmehri, S. , "Confidence regions for the location of response surface optima: the R package OptimaRegion".