Premium
Outlier detection in chemical data by fractal analysis
Author(s) -
Cramer Jeffrey A.,
Shah Sitaram S.,
Battaglia Tina M.,
Banerji Soame N.,
Obando Louis A.,
Booksh Karl S.
Publication year - 2004
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.875
Subject(s) - outlier , anomaly detection , data set , fractal dimension , pattern recognition (psychology) , dimension (graph theory) , fractal , artificial intelligence , mathematics , computer science , set (abstract data type) , bilinear interpolation , fractal analysis , latent variable , data mining , statistics , mathematical analysis , pure mathematics , programming language
Abstract A new outlier detection technique has been created which functions effectively with non‐bilinear data, a situation in which more common detection techniques have difficulty. The method involves the calculation of the data set's fractal dimension in the latent variable score space. For batch data sets a full cross‐validation is used to find outliers, while process data are analyzed using a training set. If a spectrum in either case significantly changes the fractal dimension statistically, then it is identified as an outlier. The technique shows good results for surface plasmon resonance data sets and has been shown to be effective in detecting even subtle outliers. Copyright © 2004 John Wiley & Sons, Ltd.