Robust bootstrap inference for linear models with non i.i.d. errors or dependent observations

01 January 2011 → 31 December 2014
Research Foundation - Flanders (FWO)
Research disciplines
  • Natural sciences
    • Applied mathematics in specific fields
    • Statistics and numerical methods
regression dependence robustness skewness bootstrap hederscedasticity
Project description

Develop fast robust bootstrap inference for robust regression with dependent observations or heteroscedastic or skew errors. Develop more efficient robust estimators such as S-estimators for these more complex linear models by modeling explicitly the heteroscedasticity or skewness, and investigate the properties of the resulting estimators. Develop fast and robust bootstrap inference for these new robust estimators.