Insights into the area under the receiver operating characteristic curve (AUC) as a discrimination measure in species distribution modelling

Global Ecology and Biogeography - Tập 21 Số 4 - Trang 498-507 - 2012
Alberto Jiménez‐Valverde1
1Department of Animal Biology, Faculty of Sciences, University of Málaga, 29071 Málaga, Spain and Azorean Biodiversity Group, University of Azores, Angra do Heroísmo, Portugal

Tóm tắt

ABSTRACTAim  The area under the receiver operating characteristic (ROC) curve (AUC) is a widely used statistic for assessing the discriminatory capacity of species distribution models. Here, I used simulated data to examine the interdependence of the AUC and classical discrimination measures (sensitivity and specificity) derived for the application of a threshold. I shall further exemplify with simulated data the implications of using the AUC to evaluate potential versus realized distribution models.Innovation  After applying the threshold that makes sensitivity and specificity equal, a strong relationship between the AUC and these two measures was found. This result is corroborated with real data. On the other hand, the AUC penalizes the models that estimate potential distributions (the regions where the species could survive and reproduce due to the existence of suitable environmental conditions), and favours those that estimate realized distributions (the regions where the species actually lives).Main conclusions  Firstly, the independence of the AUC from the threshold selection may be irrelevant in practice. This result also emphasizes the fact that the AUC assumes nothing about the relative costs of errors of omission and commission. However, in most real situations this premise may not be optimal. Measures derived from a contingency table for different cost ratio scenarios, together with the ROC curve, may be more informative than reporting just a single AUC value. Secondly, the AUC is only truly informative when there are true instances of absence available and the objective is the estimation of the realized distribution. When the potential distribution is the goal of the research, the AUC is not an appropriate performance measure because the weight of commission errors is much lower than that of omission errors.

Từ khóa


Tài liệu tham khảo

10.1016/S0031-3203(98)00154-X

10.1016/S0010-4825(99)00025-6

10.1111/j.1365-2664.2006.01214.x

10.1016/S0304-3800(02)00349-6

10.1016/j.ecolmodel.2011.02.011

Busby J.R., 1991, Nature conservation: cost effective biological surveys and data analysis, 64

10.1177/001316446002000104

10.1146/annurev.ecolsys.110308.120159

10.1111/j.2006.0906-7590.04596.x

10.1016/S0304-3800(02)00327-7

10.1016/j.patrec.2005.10.010

10.1017/S0376892997000088

Freeman E.(2009)PresenceAbsence: presence–absence model evaluation. R package version 1.1.3. Available at:http://www.R‐project.org(accessed March 2009)

10.1016/j.ecolmodel.2008.05.015

10.1002/joc.1276

10.1177/0272989X9101100204

10.1016/j.actao.2007.02.001

10.1111/j.1472-4642.2008.00496.x

10.1556/ComEc.10.2009.2.9

10.5735/086.046.0606

10.1111/j.1365-2699.2010.02465.x

10.1016/j.jclinepi.2007.10.011

10.1201/9781439800225

10.1111/j.1466-8238.2007.00358.x

10.1111/j.1600-0587.2009.06039.x

10.1007/978-1-4899-3242-6

Manly B.F.J., 2002, Resource selection by animals: statistical design and analysis for field studies

10.1080/01621459.2000.10473930

10.1002/bimj.200410133

10.1093/aje/kwj063

10.17161/bi.v3i0.29

10.1016/j.ecolmodel.2007.11.008

10.1016/j.ecolmodel.2005.03.026

R Development Core Team, 2008, R: a language and environment for statistical computing

10.1111/j.1466-8238.2010.00581.x

10.1177/096228029900800203

10.1111/j.1600-0587.2009.06074.x

10.1073/pnas.0901637106

10.1080/136588199241391

10.1016/0895-4356(88)90031-5

10.1016/j.apmr.2003.12.002

10.1111/j.1365-2664.2005.01052.x

10.1111/j.1541-0420.2008.01116.x

10.1214/10-AOAS331

10.5670/oceanog.2003.42

10.1093/clinchem/39.4.561

Thông tin
Thông tin xuất bản

Global Ecology and Biogeography

Tập 21 Số 4

498-507

Thông tin tác giả