Testing for no effect in nonparametric regression via spline smoothing techniques

Optimal designs for testing the function

Optimal designs for testing the functional form of a regression via nonparametric estimation techniques

✍ Stefanie Biedermann; Holger Dette 📂 Article 📅 2001 🏛 Elsevier Science 🌐 English ⚖ 126 KB

For the problem of checking linearity in a heteroscedastic nonparametric regression model under a ÿxed design assumption, we study maximin designs which maximize the minimum power of a nonparametric test over a broad class of alternatives from the assumed linear regression model. It is demonstrated