UNEQ is a parametric probabilisic class-modelling technique that makes use of the Hotelling's T2 distribution and can be considered as the modelling version of quadratic discriminant analysis. The class of interest is described by an elliptical space built around the barycentre of training data points of the class, namely the centroid vector. Orientation and eccentricity of the elliptical class space, respectively, describe correlation between the variables and their dispersion, while the width of the class space is determined according to the critical value of Hotelling’s T2 statistics at a pre-determined confidence level.
Student. The probable error of a mean. Biometrika 1908; 6: 1–25.
2.
HotellingHMultivariate quality control illustrated by air testing of sample bombsights. In: EisenhartCHastayMWWallisWA (eds). Sel. Tech. Stat. Anal. Sci. Ind. Res. Prod. Manag. Eng, New York and London: McGraw-Hill Book Company, Inc., 1947, pp. 111–184.
3.
DerdeMPMassartDL. UNEQ: a disjoint modelling technique for pattern recognition based on normal distribution. Anal Chim Acta1986; 184: 33–51.
4.
DerdeMPMassartDL. UNEQ: a class modelling supervised pattern recognition technique. Mikrochim Acta1986; 89: 139–152.
5.
OliveriPDowneyG. Multivariate class modeling for the verification of food-authenticity claims. Trends Anal Chem2012; 35: 74–86.
6.
ForinaMLanteriSSarabiaL. Distance and class space in the UNEQ class-modeling technique. J Chemom1995; 9: 69–89.
