Doctorant sous la direction de P. BONDON

Titre de la thèse: Analysis of multivariate time series with missing data : application to data of air quality and morbidity caused by pollution
Résumé de la thèse: Often, economic, hydrological, climatological and air pollution time series contain missing observations in one or more series. This results in a difficulty to the statistical analysis of the data. Missing observations can be caused by a variety of problems, such as failure of observation data on weekends, failure of the measuring equipment, an error of the researcher, delete of an outlier observation, etc. This project aims to propose a new methodology for the identification and estimation of time series multivariate models with missing observations. Therefore, the estimators of the autocorrelation function studied by Parzen(1963) and Yajima and Nishino (1999) will be extended to the multivariate context and its properties such as consistency, addiction, asymptotic distribution, etc, will be derived.