AUK: a simple alternative to the AUC

U. Kaymak, A. Ben-David, R. Potharst

Research output: Book/ReportReportAcademic

4 Downloads (Pure)

Abstract

The area under Receiver Operating Characteristic (ROC) curve, also known as the AUC-index, is commonly used for ranking the performance of data mining models. The AUC has many merits, such as objectivity and ease of interpretation. However, since it is class indifferent, its usefulness while dealing with highly skewed data sets is questionable, to say the least. In this paper, we propose a simple alternative scalar measure to the AUCindex, the Area Under an Kappa curve (AUK). The proposed AUK-index compensates for the above basic flaw of the AUC by being sensitive to the class distribution. Therefore it is particularly suitable for measuring classifiers’ performance on skewed data sets. After introducing the AUK we explore its mathematical relationship with the AUC and show that there is a nonlinear relation between them.
Original languageEnglish
Place of PublicationRotterdam
PublisherErasmus Universiteit Rotterdam
Number of pages19
Publication statusPublished - 2010

Publication series

NameERIM report series research in management
VolumeERS-2010-024-LIS

Fingerprint

Data mining
Classifiers
Defects

Cite this

Kaymak, U., Ben-David, A., & Potharst, R. (2010). AUK: a simple alternative to the AUC. (ERIM report series research in management; Vol. ERS-2010-024-LIS). Rotterdam: Erasmus Universiteit Rotterdam.
Kaymak, U. ; Ben-David, A. ; Potharst, R. / AUK: a simple alternative to the AUC. Rotterdam : Erasmus Universiteit Rotterdam, 2010. 19 p. (ERIM report series research in management).
@book{5ae22420b953415681bbb25877b6fdd9,
title = "AUK: a simple alternative to the AUC",
abstract = "The area under Receiver Operating Characteristic (ROC) curve, also known as the AUC-index, is commonly used for ranking the performance of data mining models. The AUC has many merits, such as objectivity and ease of interpretation. However, since it is class indifferent, its usefulness while dealing with highly skewed data sets is questionable, to say the least. In this paper, we propose a simple alternative scalar measure to the AUCindex, the Area Under an Kappa curve (AUK). The proposed AUK-index compensates for the above basic flaw of the AUC by being sensitive to the class distribution. Therefore it is particularly suitable for measuring classifiers’ performance on skewed data sets. After introducing the AUK we explore its mathematical relationship with the AUC and show that there is a nonlinear relation between them.",
author = "U. Kaymak and A. Ben-David and R. Potharst",
year = "2010",
language = "English",
series = "ERIM report series research in management",
publisher = "Erasmus Universiteit Rotterdam",
address = "Netherlands",

}

Kaymak, U, Ben-David, A & Potharst, R 2010, AUK: a simple alternative to the AUC. ERIM report series research in management, vol. ERS-2010-024-LIS, Erasmus Universiteit Rotterdam, Rotterdam.

AUK: a simple alternative to the AUC. / Kaymak, U.; Ben-David, A.; Potharst, R.

Rotterdam : Erasmus Universiteit Rotterdam, 2010. 19 p. (ERIM report series research in management; Vol. ERS-2010-024-LIS).

Research output: Book/ReportReportAcademic

TY - BOOK

T1 - AUK: a simple alternative to the AUC

AU - Kaymak, U.

AU - Ben-David, A.

AU - Potharst, R.

PY - 2010

Y1 - 2010

N2 - The area under Receiver Operating Characteristic (ROC) curve, also known as the AUC-index, is commonly used for ranking the performance of data mining models. The AUC has many merits, such as objectivity and ease of interpretation. However, since it is class indifferent, its usefulness while dealing with highly skewed data sets is questionable, to say the least. In this paper, we propose a simple alternative scalar measure to the AUCindex, the Area Under an Kappa curve (AUK). The proposed AUK-index compensates for the above basic flaw of the AUC by being sensitive to the class distribution. Therefore it is particularly suitable for measuring classifiers’ performance on skewed data sets. After introducing the AUK we explore its mathematical relationship with the AUC and show that there is a nonlinear relation between them.

AB - The area under Receiver Operating Characteristic (ROC) curve, also known as the AUC-index, is commonly used for ranking the performance of data mining models. The AUC has many merits, such as objectivity and ease of interpretation. However, since it is class indifferent, its usefulness while dealing with highly skewed data sets is questionable, to say the least. In this paper, we propose a simple alternative scalar measure to the AUCindex, the Area Under an Kappa curve (AUK). The proposed AUK-index compensates for the above basic flaw of the AUC by being sensitive to the class distribution. Therefore it is particularly suitable for measuring classifiers’ performance on skewed data sets. After introducing the AUK we explore its mathematical relationship with the AUC and show that there is a nonlinear relation between them.

M3 - Report

T3 - ERIM report series research in management

BT - AUK: a simple alternative to the AUC

PB - Erasmus Universiteit Rotterdam

CY - Rotterdam

ER -

Kaymak U, Ben-David A, Potharst R. AUK: a simple alternative to the AUC. Rotterdam: Erasmus Universiteit Rotterdam, 2010. 19 p. (ERIM report series research in management).