AndreaCappozzo/raedda: Robust Adaptive Eigenvalue Decomposition Discriminant Analysis

Model-based framework for classification that jointly accounts for outliers, label noise and unobserved classes in the test set, employing MVN mixture model with Parsimonious structure. It is a robust generalization of the AMDA methodology (Bouveyron, 2014) that accounts for outliers and label noise detecting observations with the lowest contributions to the overall likelihood employing impartial trimming.

install.packages("remotes")
remotes::install_github("AndreaCappozzo/raedda")