Events and Meetings of Italian Statistical Society, Statistics and Demography: the Legacy of Corrado Gini

Font Size: 
Old solutions for new problems: one-class classification based on transvariation probability
Francesca Fortunato, Angela Montanari

Last modified: 2015-09-05

Abstract


One class classification is justified when objects from one class only, the target class, are available. This is different from the traditional classification problem, which tries to distinguish between two or more classes with the training set containing objects from all of them. We propose a method for one-class classifi- cation based on Gini’s transvariation probability between a group and a constant. We explore one-class classification application in recognizing adulterated Olive Oil whereas only the spectrum of Pure Olive Oil is well-known. The method is em- ployed on dimensionally reduced data through Sparse Principal Component analysis.


Full Text: PDF