MINIMUM-PENALIZED DISTANCE ESTIMATORS OF A DENSITYAND ITS DERIVATIVES
Lesław Gajek
Abstract: Let be an -times differentiable distribution function (df)
generating the data. Let be the density of Let denote the empirical
df. The paper concerns fitting an -times differentiable function to
the data by minimizing where
denotes the -norm and is a sequence of smoothing parameters.
Let be an (approximate) minimizer of the above problem. We establish an
upper bound for with respect to the choice of
In particular, the choice of results in the optimal -rate of
convergence of to The estimation is also evaluated.