An introduction to ROC analysis

By Tom Fawcett. 2006

21,593 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Health Information Management, Information Systems, Statistics and Probability, Evaluation & Benchmarks

How researchers see this paper

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

In Artificial Intelligence

In Health Information Management

In Information Systems

In Statistics and Probability

In Evaluation & Benchmarks

Citations in the corpus

Deep Interest Network for Click-Through Rate Prediction (2017, methodology): “In CTR prediction field, AUC is a widely used metric[8].”
Early Methods for Detecting Adversarial Images (2016, background): “Moreover, the AUROC can be interpreted as the probability that an adversarial example has a greater detector score/value than a clean example (Fawcett, 2005).”
Opening black box Data Mining models using Sensitivity Analysis (2011, methodology): “When modeling class probabilities (case iii ), we adopt the global Area Under the Curve (AUC), which weights the area of each ROC class according to its prevalence in the data [24].”
Multi-Fidelity Automatic Hyper-Parameter Tuning via Transfer Series Expansion (2019, neutral): “The evaluation criterion of experiments is AUC score (Fawcett 2006).”
Modeling Consumer Preferences and Price Sensitivities from Large-Scale Grocery Shopping Transaction Logs (2017, methodology): “Therefore we apply the area under the curve (AUC) metric to evaluate the performance of category purchase prediction, which is suited to imbalanced binary prediction tasks [15].”
Enhancing The Reliability of Out-of-distribution Image Detection in Neural Networks (2017, background): “The AUROC can be interpreted as the probability that a positive example is assigned a higher detection score than a negative example (Fawcett, 2006).”
A data-driven approach to predict the success of bank telemarketing (2014, background): “…from a probabilistic outcome by assigning a threshold The receiver operating charac-9 teristic (ROC) curve shows the performance of a two class classiﬁer across the range of possible threshold ( D ) values, plotting one minus the speciﬁcity ( x -axis) versus the sensitivity ( y -axis) [11].”
Ad click prediction: a view from the trenches (2013, methodology): “We compute metrics such as AucLoss (that is, 1 − AUC, where AUC is the standard area under the ROC curve metric [13]), LogLoss (see Eq.”
pROC: an open-source package for R and S+ to analyze and compare ROC curves (2011, methodology): “AUCs are computed with trapezoids [4].”
The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets (2015, methodology): “Here, we review the theoretical background of commonly used evaluation measures, specifically ROC [24, 25], PRC [26], concentrated ROC (CROC) [27], and Cost Curves (CC) [28].”
Towards the Deployment of Machine Learning Solutions in Network Traffic Classification: A Systematic Survey (2019, positive): “Particularly, the ROC curve is obtained by computing the sensitivity and specificity varying the classifier’s discrimination threshold [42].”
Network Inference and Biological Dynamics (2011, methodology): “The performance of each inference scheme is quantiﬁed by the area under the receiver operating characteristic (ROC) curve (AUR), averaged over 20 data sets [Fawcett (2005)].”
Generalized Out-of-Distribution Detection: A Survey (2021, neutral): “Evaluation Apart from F-scores, AUROC, and AUPR, another commonly-used metric is FPR@TPR x , which measures the FPR when the TPR is x (e.g.,0.95).”
Weighted extreme learning machine for imbalance learning (2013, methodology): “Another interesting tool, receiver operating characteristics (ROC) graph [16], provides a visual illustration of the performance of classifiers on binary datasets, where a classifier corresponds to a point.”
Multiclass Real-Time Intent Recognition of a Powered Lower Limb Prosthesis (2009, methodology): “In order to find the best classifier for each frame length, the area under the receiver operator characteristics (AUC) curve [29] was used as the performance metric.”
Semi-Supervised Novelty Detection (2010, background): “These should be considered well-known (see for example the paper of Fawcett, 2006,and references therein, for an overview of ROC analysis over sets of classifiers).”
A Dataset for Breast Cancer Histopathological Image Classification (2016, background): “If the proportion of positive to negative instances changes in a test set, the ROC curves will not change [34].”
Deep learning of the tissue-regulated splicing code (2014, background): “We will refer to this as the ‘decrease, no change, increase’ (DNI) code, with targets yDNIk ðx; ti; tjÞ.”