|
A Feature Selection Method for Imbalance
|
|
|
Ilnaz Jamali
Department of Electrical and Computer Engineering
Shiraz University
Shiraz, Iran
ilnazjamali@yahoo.com
|
|
Sattar Hashemi
Department of Electrical and Computer Engineering
Shiraz University
Shiraz, Iran
s_hashemi@shirazu.ac.ir
|
|
|
|
|
|
Abstract
.This paper introduces a game theoretic framework for feature selection in imbalance data sets. In this method which is called FSSH (Feature Selection based on Shapley value), first some coalitions will be constructed and the marginal importance of each feature in its coalition will be computed. Then, the weighted mean of each feature’s value considered as the Shapley value. Finally features will be ranked according to their Shapley value and high ranked features will be selected in the realm of feature selection. Experimental results and comparison with several existing feature selection methods show the advantages of presented approach across the data sets adopted in this study.
|
|
|
|
Keywords
:
feature selection, imbalance data sets, game theor
|
|
|
|
URL: http://dx.doi.org/10.7321/jscse.v3.n3.89
|
|
|
|
|