Estimación no paramétrica de curvas notables para datos dependientes
ISSN: 0213-8190
Year of publication: 1989
Volume: 4
Issue: 2
Pages: 69-88
Type: Article
More publications in: Trabajos de estadística
Sustainable development goals
Abstract
Let {Xt: t Î Z} be a stationary valued in Rp time series, verifying the condition a-mixing or L2-stability. For a sample of size n, a general class of estimators of the density function f(x) associated to the process and of the autoregression function in order k is defined: r(y) = E(g(Xt-k+1 ... Xt) = y), y Î Rk being g a real function. The following asymptotic properties of these estimators are studied: punctual consistency (almost sure and in Ln); uniform consistency almost sure; bias, variance and asymptotic normality