Pilar García Soidán , Manuel Febrero Bande , Wenceslao González Manteiga
In this work, the asymptotic normality of the kernel variogram estimator is established, which provides confidence intervals for the variogram via the normal quantiles. A different alternative is proposed by using the Bootstrap techniques. In particular, we will consider a gaussian process where the second order structure is given by the variogram estimator; then, the resampling will be done from this random process and proved to be consistent. In consequence, the quantiles of the variogram estimator may be approximated from those of its Bootstrap counterpart.
© 2008-2024 Fundación Dialnet · Todos los derechos reservados