Abstract |
: |
We derive an approximation of a density estimator based on weakly dependent random vectors by a density estimator built from independent random vectors. We construct, on a sufficiently rich probability space, such a pairing of the random variables of both experiments that the set of observations (X-1,..., X-n) from the time series model is nearly the same as the set of observations (Y-1,..., Y-n) from the i.i.d. model. With a high probability, all sets of the form ((X-1,..., X-n)Delta(Y-1,..., Y-n))boolean AND([a(1), b(1)] x...x [a(d), b(d)]) contain no more than O(([n(1/2) Pi(b(i) - a(i))] + 1)log(n)) elements, respectively. Although this does not imply very much for parametric problems, it has important implications in nonparametric statistics. It yields a strong approximation of a kernel estimator of the stationary density by a kernel density estimator in the i.i.d. model. Moreover, it is shown that such a strong approximation is also valid for the standard bootstrap and the smoothed bootstrap. Using these results we derive simultaneous confidence bands as well as supremum-type nonparametric tests based on reasoning for the i.i.d. model. |