1. A Darling-Erdős-type CUSUM-procedure for functional data.
- Author
-
Torgovitski, Leonid
- Subjects
CUSUM technique ,MATHEMATICS theorems ,COMPUTER simulation ,TEMPERATURE ,HYPOTHESIS - Abstract
The focus of the paper is nonparametric detection of changes in the mean of $$m$$ -dependent stationary functional data via a cumulative sum (CUSUM) procedure. We consider a projection-based quasi-maximum likelihood CUSUM-procedure which relies on a Darling-Erdős-type limit theorem. Under mild moment assumptions we investigate the asymptotic properties under the null hypothesis and show consistency under the alternatives of either an abrupt or a gradual change in the mean. The finite sample behavior is illustrated in a small simulation study including an application to temperature data from Hohenpeißenberg (Bavaria, Germany). [ABSTRACT FROM AUTHOR]
- Published
- 2015
- Full Text
- View/download PDF