Infinitesimally Robust Estimation in General Smoothly Parametrized Models

TL;DR

This paper discusses a robust estimation method for smoothly parametrized models, emphasizing exponential families, and introduces object-oriented implementation for optimal robustness, validated through real data analysis using R packages.

Contribution

It presents a general approach to robust estimation in smooth models with object-oriented implementation and practical validation with R packages.

Findings

Effective robust estimators for exponential families

Successful application to real datasets

Implementation via R packages ROptEst and RobLox

Abstract

We describe the shrinking neighborhood approach of Robust Statistics, which applies to general smoothly parametrized models, especially, exponential families. Equal generality is achieved by object oriented implementation of the optimally robust estimators. We evaluate the estimates on real datasets from literature by means of our R packages ROptEst and RobLox.