Beyond Accuracy, F-Score and ROC: A Family of Discriminant Measures for Performance Evaluation

By Marina Sokolova, Nathalie Japkowicz, Stan Szpakowicz. Australian Conference on Artificial Intelligence, 2006

1,831 citations recorded by the source corpus.

Subfields: Accounting, Artificial Intelligence

View Paper: https://www.semanticscholar.org/paper/e3d8a9d35232d7967780019f0fa93626ca1373cb

How researchers see this paper

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

In Accounting

In Artificial Intelligence

Citations in the corpus

Recent Advances in Open Set Recognition: A Survey (2018, background): “Youden’s index can express an algorithm’s ability to avoid failure [146], and it is bounded in [−1, 1], where higher value indicates an algorithm more resistant to failure.”
Segmentation using Deformable Spatial Priors with Application to Clothing (2010, positive): “We use Balanced-Accuracy [21] for the correct labelling compared to the ground truth segmentations.”
Efficient Water Quality Prediction Using Supervised Machine Learning (2019, methodology): “Precision is calculated with the formula shown in Equation (10), where TP refers to true positive and FP refers to false positive [7,35,36].”
Classification assessment methods (2020, methodology): “(9) [20].”
Identification of MCI Individuals Using Structural and Functional Connectivity Networks (2011, background): “The Youden’s index (Y), Balanced ACcuracy (BAC) and F-score (F) are defined respectively as [27]”
Evaluation of disorder predictions in CASP9 (2011, result): “42,43 Calculations on CASP data have shown a high correlation of this measure with the MCC (Spearman’s q 5 0.9), and therefore these results are not shown.”
An overview of ensemble methods for binary classifiers in multi-class problems: Experimental study on one-vs-one and one-vs-all schemes (2011, background): “However, in the literature most of the performance measures are designed only for two-class problems [69] [27].”
The evaluation of Network Anomaly Detection Systems: Statistical analysis of the UNSW-NB15 data set and the comparison with the KDD99 data set (2016, background): “The accuracy (So-In et al., 2014; Sokolova et al., 2006) is the rate of the correctly classified records to all the records, whether correctly or incorrectly classified, which is denoted as The false alarm rate (FAR) is the average ratio of the misclassified to classified records either normal or…”
An Intelligent Fault Diagnosis Method Using Unsupervised Feature Learning Towards Mechanical Big Data (2016, methodology): “To compare their performances with that of the proposed method, F-score is adopted [40], which is a commonly used criterion measuring the performance of a classification method.”
On the effectiveness of preprocessing methods when dealing with different levels of class imbalance (2012, methodology): “Chen et al. [43] introduced a method based on pruning support vectors of the majority class.”
Feature selection for high-dimensional class-imbalanced data sets using Support Vector Machines (2014, methodology): “Another metric commonly used to assess classiﬁcation models is the Area under the curve (AUC), see [30].”
Individual tree detection from Unmanned Aerial Vehicle (UAV) derived canopy height model in an open canopy mixed conifer forest (2017, neutral): “[66], using the following equations [67,68]:”