Estimating the number of change-points via Schwarz' criterion (Q1101154): Difference between revisions

This short paper is a contribution to the problem of estimating the number of change-points. The model consists of independent, normally distributed observations with common variance but mean jumps at unknown times. For a fixed number of changes, all the parameters are classically estimated by the likelihood method; then, the estimated likelihood is penalized as in Schwarz's criterion [see \textit{G. Schwarz}, Ann. Stat. 6, 461-464 (1978; Zbl 0379.62005)] to give the estimated number of changes. In this context: normality and asymptotically well-separated change times, the estimator is proved to be consistent.