Two flexible models obtained by perturbation of symmetric densities are considered: the Skew Exponential Power distribution and the Skew t distribution. They are able to fit the distribution of data which show either skewness or non-normal tails and provide an alternative to robust procedures when deviations from normality occur. The benefits of the adoption of these models for handling non¬-normal data are discussed and their application in the robustness context is shown in two examples.
|Titolo:||Flexible models and robustness|
|Data di pubblicazione:||2009|
|Appare nelle tipologie:||4.1 Contributo in Atti di convegno|