Kernel spatial regression estimation for non-stationary process with applications

Date(s) : 12/06/2017   iCal
15 h 30 min - 16 h 30 min

Let ( Z i , i ∈ Z N ) be a spatial process where Z i = ( X i , Y i ) are such the Y i ‘s are real-valued and integrable variable and X i ‘s are valued in a (semi-)metric separable space ( E ,d ). This work deals with the problem of the estimation the regression function, r defined by r( x ) =E( Y i | X i =x ) when the process ( Z i ) is not strictly stationary. We study the asymptotic behavior of the kernel estimator under mixing and local stationarity conditions. We also discuss the theoretical and practical aspects of relaxing the stationary hypothesis and present some applications.

Portrait vidéo

Catégories Pas de Catégories

Retour en haut 

Secured By miniOrange