Date/heure
1 décembre 2020
14:00 - 15:00
Oratrice ou orateur
Youssef Salman
Catégorie d'évènement Séminaire des doctorants
Résumé
In statistical analysis, change point detection aims to identify the times when the probability distribution of a stochastic process or a time series, or the parameter of the time series models changes. In general, the problem concerns both detecting the changes and identifying their locations. My goal is not only to detect the big breakpoint, but also, the detection of the small changes. The likelihood ratio test is used to detect these changes (small and big changes). The distribution
under the null and the alternatives hypothesis of the test was did by the LAN property (Locally asymptotic normal) and the Le Cam’s third lemma. The optimality of the test was proved at the end of the job.