A new method for estimation and model selection: \(\rho\)-estimation (Q510164)

In this paper the authors discuss a new estimation procedure which leads to ``a new estimation procedure and can be applied in various statistical frameworks including density and regression\dots''. In the regression framework where the errors have rather unusual distributions, the least squares method is far from optimal. The authors consider the form \(Z_i=f(w_i)+\epsilon_i\) for \(1\leq i\leq n\), where \(Z_i\) are real observations and the \(\epsilon_i\) i.i.d. errors with density \(p\) with respect to the Lebesque measure. In case of density estimation ``they asymptotically coincide with the celebrated maximum likelihood estimators at least when the statistical model is regular enough and contains the true density to estimate''.