ROCR: visualizing classifier performance in R.

Department of Computational Biology and Applied Algorithmics, Max-Planck-Institute for Informatics, Saarbrücken, Germany.
Bioinformatics (Impact Factor: 4.62). 11/2005; 21(20):3940-1. DOI: 10.1093/bioinformatics/bti623
Source: PubMed

ABSTRACT ROCR is a package for evaluating and visualizing the performance of scoring classifiers in the statistical language R. It features over 25 performance measures that can be freely combined to create two-dimensional performance curves. Standard methods for investigating trade-offs between specific performance measures are available within a uniform framework, including receiver operating characteristic (ROC) graphs, precision/recall plots, lift charts and cost curves. ROCR integrates tightly with R's powerful graphics capabilities, thus allowing for highly adjustable plots. Being equipped with only three commands and reasonable default values for optional parameters, ROCR combines flexibility with ease of usage. AVAILABILITY: ROCR can be used under the terms of the GNU General Public License. Running within R, it is platform-independent. CONTACT:

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We examined a secondary contact zone between two species of desert tortoise, Gopherus agassizii and G. morafkai. The taxa were isolated from a common ancestor during the formation of the Colorado River (4–8 mya) and are a classic example of allopatric speciation. However, an anomalous population of G. agassizii comes into secondary contact with G. morafkai east of the Colorado River in the Black Mountains of Arizona and provides an opportunity to examine reinforcement of species' boundaries under natural conditions. We sampled 234 tortoises representing G. agassizii in California (n = 103), G. morafkai in Arizona (n = 78), and 53 individuals of undetermined assignment in the contact zone including and surrounding the Black Mountains. We genotyped individuals for 25 STR loci and determined maternal lineage using mtDNA sequence data. We performed multilo-cus genetic clustering analyses and used multiple statistical methods to detect levels of hybridization. We tested hypotheses about habitat use between G. agassizii and G. morafkai in the region where they co-occur using habitat suitability models. Gopherus agassizii and G. morafkai maintain independent taxonomic identities likely due to ecological niche partitioning, and the maintenance of the hybrid zone is best described by a geographical selection gradient model.
    Ecology and Evolution 04/2015; 5(10). DOI:10.1002/ece3.1500 · 1.66 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: We explored the seasonal potential fishing grounds of neon flying squid (Ommastrephes bartramii) in the western and central North Pacific using maximum entropy (MaxEnt) models fitted with squid fishery data as response and environmental factors from remotely sensed [sea surface temperature (SST), sea surface height (SSH), eddy kinetic energy (EKE), wind stress curl (WSC) and numerical model-derived sea surface salinity (SSS)] covariates. The potential squid fishing grounds from January–February (winter) and June–July (summer) 2001–2004 were simulated separately and covered the near-coast (winter) and offshore (summer) forage areas off the Kuroshio–Oyashio transition and subarctic frontal zones. The oceanographic conditions differed between regions and were regulated by the inherent seasonal variability and prevailing basin dynamics. The seasonal and spatial extents of potential squid fishing grounds were largely explained by SST (7–17°C in the winter and 11–18°C in the summer) and SSS (33.8–34.8 in the winter and 33.7–34.3 in the summer). These ocean properties are water mass tracers and define the boundaries of the North Pacific hydrographic provinces. Mesoscale variability in the upper ocean inferred from SSH and EKE were also influential to squid potential fishing grounds and are presumably linked to the augmented primary productivity from nutrient enhancement and entrainment of passive plankton. WSC, however, has the least model contribution to squid potential fishing habitat relative to the other environmental factors examined. Findings of this work underpin the importance of SST and SSS as robust predictors of the seasonal squid potential fishing grounds in the western and central North Pacific and highlight MaxEnt's potential for operational fishery application.
    Fisheries Oceanography 02/2015; 24(2):190-203. DOI:10.1111/fog.12102 · 2.54 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The Brook Trout Salvelinus fontinalis is an important species of conservation concern in the eastern USA. We developed a model to predict Brook Trout population status within individual stream reaches throughout the species’ native range in the eastern USA. We utilized hierarchical logistic regression with Bayesian estimation to predict Brook Trout occurrence probability, and we allowed slopes and intercepts to vary among ecological drainage units (EDUs). Model performance was similar for 7,327 training samples and 1,832 validation samples based on the area under the receiver operating curve (∼0.78) and Cohen's kappa statistic (0.44). Predicted water temperature had a strong negative effect on Brook Trout occurrence probability at the stream reach scale and was also negatively associated with the EDU average probability of Brook Trout occurrence (i.e., EDU-specific intercepts). The effect of soil permeability was positive but decreased as EDU mean soil permeability increased. Brook Trout were less likely to occur in stream reaches surrounded by agricultural or developed land cover, and an interaction suggested that agricultural land cover also resulted in an increased sensitivity to water temperature. Our model provides a further understanding of how Brook Trout are shaped by habitat characteristics in the region and yields maps of stream-reach-scale predictions, which together can be used to support ongoing conservation and management efforts. These decision support tools can be used to identify the extent of potentially suitable habitat, estimate historic habitat losses, and prioritize conservation efforts by selecting suitable stream reaches for a given action. Future work could extend the model to account for additional landscape or habitat characteristics, include biotic interactions, or estimate potential Brook Trout responses to climate and land use changes.Received May 9, 2014; accepted August 26, 2014
    Transactions of the American Fisheries Society 01/2015; 144(1). DOI:10.1080/00028487.2014.963256 · 1.31 Impact Factor


Available from