On Rohlf's Method for the Detection of Outliers in Multivariate Data
1995
Rohlf (1975, Biometrics31, 93-101) proposed a method of detecting outliers in multivariate data by testing the largest edge of the minimum spanning tree. It is shown here that tests against the gamma distribution are extremely liberal. Furthermore, results depend on the correlation structure of the data if Euclidean distances are used. While the use of generalized distances might avoid this difficulty, the construction of the robust estimates required to carry out the test with generalized distances provides in itself information on outliers which leaves Rohlf?s procedure superfluous. It is concluded that Rohlf?s method does not provide a useful formal test.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
24
References
17
Citations
NaN
KQI