Strong uniform convergence rates in robust nonparametric time series analysis and prediction: Kernel regression estimation from dependent observations (Q1088357)

Let \(\{Z_ t\}\) be a strictly stationary time series. The authors introduce a predictor for \(Z_{N+1}\) based on \(Z_ 1,...,Z_ N\) as a functional M-estimator connected with \({\mathcal L}(Z_{p+1}| Z_ 1,...,Z_ p)\) when \(Z_ t\) is Markovian of order p. Its strong uniform convergence rate (s.u.c.r.) is given. Then robust regression kernel estimators from identically distributed and \(\phi\)-mixing random pairs \(\{(X_ i,Y_ i)\); \(i=1,...,n\}\) are investigated and s.u.c.r. for estimators of \(E(Y_ 1| X_ 1=.)\) and of the probability density of \(X_ 1\) are obtained.