DMCA
Improving Identification of Difficult Small Classes by Balancing Class Distribution [ftp://ftp.cs.uta.fi/pub/reports/pdf/A-2001-2.pdf (2001)
| Citations: | 61 - 0 self |
BibTeX
@TECHREPORT{Laurikkala01improvingidentification,
author = {Jorma Laurikkala},
title = {Improving Identification of Difficult Small Classes by Balancing Class Distribution [ftp://ftp.cs.uta.fi/pub/reports/pdf/A-2001-2.pdf},
institution = {},
year = {2001}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
Abstract. We studied three methods to improve identification of difficult small classes by balancing imbalanced class distribution with data reduction. The new method, neighborhood cleaning rule (NCL), outperformed simple random and one-sided selection methods in experiments with ten data sets. All reduction methods improved identification of small classes (20-30%), but the differences were insignificant. However, significant differences in accuracies, true-positive rates and true-negative rates obtained with the 3-nearest neighbor method and C4.5 from the reduced data favored NCL. The results suggest that NCL is a useful method for improving the modeling of difficult small classes, and for building classifiers to identify these classes from the real-world data. 1
Keyphrases
difficult small class fi pub report class distribution ftp significant difference simple random new method ten data set one-sided selection method true-negative rate data reduction neighborhood cleaning rule true-positive rate real-world data reduction method useful method imbalanced class distribution 3-nearest neighbor method small class
