We consider a nonlinear polynomial regression model in which we wish to test the null hypothesis of structural stability in the regression parameters against the alternative of a break at an unknown time. We derive the extreme value distribution of a maximum-type test statistic which is asymptotically equivalent to the maximally selected likelihood ratio. The resulting test is easy to apply and has good size and power, even in small samples.
Publié le : 2008-08-15
Classification:
change-point analysis,
extreme value asymptotics,
Gaussian processes,
Legendre polynomials,
linear regression,
polynomial regression
@article{1219669624,
author = {Aue, Alexander and Horv\'ath, Lajos and Hu\v skov\'a, Marie and Kokoszka, Piotr},
title = {Testing for changes in polynomial regression},
journal = {Bernoulli},
volume = {14},
number = {1},
year = {2008},
pages = { 637-660},
language = {en},
url = {http://dml.mathdoc.fr/item/1219669624}
}
Aue, Alexander; Horváth, Lajos; Hušková, Marie; Kokoszka, Piotr. Testing for changes in polynomial regression. Bernoulli, Tome 14 (2008) no. 1, pp. 637-660. http://gdmltest.u-ga.fr/item/1219669624/