All Rights Reserved
AccessEcon LLC 2006, 2008.
Powered by MinhViet JSC

 
Téa Ouraga
 
''A note on Gini Principal Component Analysis''
( 2019, Vol. 39 No.2 )
 
 
In this paper, a principal component analysis based on the Gini index - Gini PCA - is proposed in order to deal with contaminated samples. The operator underlying the Gini index is a covariance-based operator, which provides a l1 metric well suited for dealing with outliers. It is shown, with simple Monte Carlo experiments, that the results of the standard Principal Component Analysis (PCA) may be drastically aff ected whereas some robustness holds with Gini PCA.
 
 
Keywords: Gini, PCA, Robutsness
JEL: C1 - Econometric and Statistical Methods: General
C4 - Econometric and Statistical Methods: Special Topics
 
Manuscript Received : Mar 07 2019 Manuscript Accepted : May 02 2019

  This abstract has been downloaded 1121 times                The Full PDF of this paper has been downloaded 166339 times