ROC Graphs: Notes and Practical Considerations for Researchers

By Tom Fawcett. 2007

2,212 citations recorded by the source corpus.

View Paper: https://www.semanticscholar.org/paper/44bb1605c4ab8a8ce2764fa20424f6a148101ca4

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Measuring the Effectiveness of Various Features for Thematic Information Extraction From Very High Resolution Remote Sensing Imagery (2015, methodology): “Many metrics in the remote sensing field are generally used to evaluate results of the thematic information extraction [35], [36], including classification accuracy of the specific classes, Kappa Index of Agreement (KIA) [37], and area under an ROC Curve (AUC) [38].”
Do We Know Who Will Drop Out?: A Review of the Predictors of Dropping out of High School: Precision, Sensitivity, and Specificity (2013, methodology): “Following the recommendations of signal detection theory (Fawcett, 2004; Hanley & McNeil, 1982; Swets, 1988; Vivo & Franco, 2008; Zwieg & Campbell, 1993), the ROC plot in Figure 2 plots the true-positive proportion against the false-positive proportion.”
Local binary patterns variants as texture descriptors for medical image analysis (2010, methodology): “The area under the ROC-curve (AUC) is a scalar measure of classification performance, which can be interpreted as the probability that the classifier will assign a higher score to a randomly picked positive sample than to a randomly picked negative sample.”
DiBS: Differentiable Bayesian Structure Learning (2021, methodology): “The area under the receiver operating characteristic curve (AUROC) evaluates proper uncertainty quantification of the posterior.”
Embedded Malware Detection Using Markov n-Grams (2008, methodology): “The value of this threshold was tuned for the best performance in the ROC space using a randomly sampled training dataset which was 5% of the total testing dataset [16].”
Information, Divergence and Risk for Binary Experiments (2009, methodology): “In this paper we draw connections between Receiver Operating Characteristic (ROC) curves, (Fawcett, 2006, 2004; Flach, 2003; Flach and Wu, 2005; Maxion and Roberts, 2004) the Area Under ROC Curve (AUC), (Cortes and Mohri, 2004; Hand, 2008; Hand and Till, 2001; Hanley and McNeil, 1982) and Cost…”
Handcrafted vs. non-handcrafted features for computer vision classification (2017, methodology): “The area under the ROC curve (AUC) [61] is used to assess performance.”
ROC analysis: applications to the classification of biological sequences and 3D structures (2008, methodology): “An ROC curve is thus a two-dimensional graph that visually depicts the relative trade-offs between the errors (false positives) and benefits (true positives) [11].”
Recommender Systems (2016, unknown): “A discussion of ROC curves in the context of classification may be found in [195], although the same ideas are also applicable to the case of recommender systems.”
Measuring classifier performance: a coherent alternative to the area under the ROC curve (2009, background): “The fact that choices of the classiﬁcation threshold t in concave intervals of the ROC curve do not minimise the loss for any choice of c means that one can produce a classiﬁer superior to that summarised by such a ROC curve (Provost and Fawcett 1997; Scott et al. 1998; Fawcett 2004).”
Current status and key issues in image steganography: A survey (2014, background): “of false positive rate versus true positive rate [6,14].”
No NAT'd User Left Behind: Fingerprinting Users behind NAT from NetFlow Records Alone (2014, methodology): “A random classiﬁer has an area of 0.5, the ideal one has an area of 1. Under this metric, we conﬁrmed that Random Forest performs better than other algorithms within our framework (SVM was the worst) [10]. Indeed, Table IV reports the average ROC Area related to 6 different users selected among the ”
HPatches: A Benchmark and Evaluation of Handcrafted and Learned Local Descriptors (2017, methodology): “However, whereas the ROC [13] is used there, AP is preferred here [30] since the dataset is highly unbalanced, with the vast majority ( 10 6 ) of patch pairs being negative.”
Local Causal and Markov Blanket Induction for Causal Discovery and Feature Selection for Classification Part I: Algorithms and Empirical Evaluation (2010, methodology): “The performance is measured by AUC (Fawcett, 2003) on binary tasks and proportion of correct classifications on multiclass tasks.”
Cybercrime detection in online communications: The experimental case of cyberbullying detection in the Twitter network (2016, background): “The area under the curve provides a signal of the discriminatory rate of the classifier at various operating points (Fawcett, 2004, 2006; Prieto et al., 2014; Provost & Fawcett, 1997).”