C4.5 and Imbalanced Data sets: Investigating the eect of sampling method, probabilistic estimate, and decision tree structure

By Nitesh V. Chawla. 2003

297 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

Hellinger distance decision trees are robust and skew-insensitive (2011, background): “Previous research has demonstrated that unpruned decision trees are more effective in their predictions on minority class, and also result in improved calibrated estimates (Provost and Domingos 2003; Chawla 2003).”
Mining with rarity: a unifying framework (2004, background): “This is partially due to the fact that these costs often depend on multiple considerations that are not easily compared [10].”
Training and assessing classification rules with imbalanced data (2012, methodology): “As mentioned by Chawla et al. (2002), generating synthetic examples, instead of simply increasing the multiplicity of the rare events, has the desirable effect of causing the classifier to identify larger decision regions associated to the minority class.”
Data Mining for Imbalanced Datasets: An Overview (2005, background): “Various studies in imbalanced datasets have used different variants of over and under sampling, and have presented (sometimes conflicting) viewpoints on usefulness of oversampling versus undersampling (Chawla, 2003; Maloof, 2003; Drurnmond and Holte, 2003; Batista et al., 2004).”
Editorial: special issue on learning from imbalanced data sets (2004, background): “First, a hint at the fact that research on the class imbalance problem is starting to mature is the fact that a big proportion of the 2003 workshop was occupied by the comparison of previously proposed schemes for dealing with the class imbalanced problem [9][14][37].”
Inverse random under sampling for class imbalance problem and its application to multi-label classification (2012, background): “Since pruning can reduce the minority class coverage in the decision trees in highly unbalanced data sets [32,9], all the results reported are without pruning.”
Cluster-based under-sampling approaches for imbalanced data distributions (2009, background): “However, many datasets in real applications involve imbalanced class distribution problem (Chawla, 2003; Chyi, 2003; Japkowicz, 2000, 2001; Jo & Japkowicz, 2004; Maloof, 2003; Zhang & Mani, 2003).”
Adaptive Machine Learning for Credit Card Fraud Detection (2015, methodology): “The authors of [55, 103, 235] recommend to leave the tree unpruned and to use Laplace smoothing for obtaining posterior probabilities.”
Using Class Imbalance Learning for Software Defect Prediction (2013, methodology): “Default parameters are used, except that we disable the tree pruning, because pruning may remove leaves describing the minority concept when data are imbalanced [46].”