Identification of Outliers

摘要: One of the most vexing problems in data analysis is determination whether or not to discard some observations because they are inconsistent with rest and/or probability distribution assumed be underlying data. direction research activity related this problem that study robust statistical procedures (cf. Huber [1981]), primarily for estimating population parameters which insensitive effect “outliers”, i.e., model random process generating observations. Typically, procedure involves “trimming” down-weighting procedure, wherein fraction extreme automatically eliminated given less weight guard against potential outliers.