Online data validation: Outlier detection and accomodation

Jan Nygaard Nielsen, Henrik Madsen

    Research output: Book/ReportReport

    Abstract

    The objective of this report is to discuss methods for identifying, rejecting, making allowances for, or minimizing the influence of, outlyer observations.The literature provides a variety of vague definitions of the concept of an outlying observation, see (Barnett and Lewis, 1994), yet consensus seems to be within reach with the following definition:An Outlier is an observation (or a subset of observations) which appear to be inconsistent with a model of the majority of the data.
    Original languageEnglish
    Number of pages27
    Publication statusPublished - 1998

    Cite this