Learning Decision Trees Using the Area Under the ROC Curve

By Cèsar Ferri, Peter A. Flach, JosÃ© Hernandez-Orallo. 2002

345 citations recorded by the source corpus.

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

Generalized and Scalable Optimal Sparse Decision Trees (2020, background): “Let us deﬁne the area under the ROC convex hull (ROCCH) (Ferri et al., 2002) for a tree.”
The Geometry of ROC Space: Understanding Machine Learning Metrics through ROC Isometrics (2003, background): “The isometrics of this AUC-split metric can be easily shown to be straight parallel lines, but the metric is skew-sensitive.”
A study of the behavior of several methods for balancing machine learning training data (2004, methodology): “For decision trees, we could use the class distributions at each leaf as a score or, as proposed in [9], by ordering the leaves by their positive class accuracy and producing several trees by re-labelling the leaves, one at a time, from all forecasting negative class to all forecasting positive clas”
Machine Learning - The Art and Science of Algorithms that Make Sense of Data (2012, background): “2 is a much more recent development (Ferri et al., 2002; Provost and Domingos, 2003).”
Using AUC and accuracy in evaluating learning algorithms (2005, methodology): “As a new measure (such as AUC) is discovered and proved to be better than a previous measure (such as accuracy), we can redesign most learning algorithms to optimize the new measure [12], [22].”
Comparing naive Bayes, decision trees, and SVM with AUC and accuracy (2003, methodology): “Recently, other researchers used AUC to construct learning algorithms[8, 15].”
Hinge Rank Loss and the Area Under the ROC Curve (2007, methodology): “Apart from that, it was experimentally observed in [3] that there was no signiﬁcant diﬀerence in AUC-scores between SVMs trained in the standard way and other approaches tailored to directly maximize the AUC, like RankBoost [17], AUCsplit (local optimization of AUC) [21], or ROC-SVM [3].”
Cost curves: An improved method for visualizing classifier performance (2006, background): “The counts at the leaves may also be modified, thus changing the leaf’s classification, allowing a single tree to produce multiple points (Bradford et al., 1998; Ferri et al., 2002).”
A support vector method for multivariate performance measures (2005, methodology): “Also, methods for optimizing ROCArea have been proposed in the area of decision trees (Ferri et al., 2002), neural networks (Yan et al.”
How to reverse-engineer quality rankings (2012, background): “2007) or give insight into how to approximately solve versions of extremely large ranking problems quickly (Tsochantaridis et al. 2005; Freund et al. 2003; Cossock and Zhang 2006; Joachims 2002; Burges et al. 2006; Xu et al. 2008; Le and Smola 2007; Ferri et al. 2002; Ataman et al. 2006).”
Class Imbalances versus Class Overlapping: An Analysis of a Learning System Behavior (2004, methodology): “We also implemented the method proposed in [3] to obtain the ROC curves and the corresponding AUCs from the standard classifiers induced by C4.”
The relationship between Precision-Recall and ROC curves (2006, methodology): “Ferri et al. (2002) alter decision trees to use the AUC-ROC as their splitting criterion, Cortes and Mohri (2003) show that the boosting algorithm Rank-Boost (Freund et al., 1998) is also well-suited to optimize the AUC-ROC, Joachims (2005) presents a generalization of Support Vector Machines which”
Introduction to the BioChemical Library (BCL): An Application-Based Open-Source Toolkit for Integrated Cheminformatics and Machine Learning in Computer-Aided Drug Discovery (2022, background): “The ﬁ rst three options rate the feature to split the dataset by information gain, Gini index, and area under the curve of the local ROC curves (Ferri et al., 2002), respectively.”
AUC-Maximized Deep Convolutional Neural Fields for Protein Sequence Labeling (2016, methodology): “In particular, (a) Cortes et al. [7] optimized AUC by RankBoost algorithm; (b) Ferri et al. [15] trained a decision tree by using AUC as splitting criteria; (c) Herschtal and Raskutti [21] trained a neural network by optimizing AUC; and (d) Joachims [23] proposed a generalized Support Vector…”
Integer optimization methods for machine learning (2012, background): “…and there are many works that discuss how to approximately solve extremely large ranking problems quickly [Freund et al., 2003a, Tsochantaridis et al., 2005, Joachims, 2002, Cossock and Zhang, 2006, Burges et al., 2006, Xu et al., 2008, Le and Smola, 2007, Ferri et al., 2002, Ataman et al., 2006].”
Subgroup Discovery with CN2-SD (2004, methodology): “A description of this method applied to decision tree induction can be found in the paper by Ferri-Ramı́rez et al. (2002).”