Cost-Sensitive Learning vs. Sampling: Which is Best for Handling Unbalanced Classes with Unequal Error Costs?
Gary M. WeissKate McCarthyBibi ZabarPublished in: DMIN (2007)
Keyphrases
- cost sensitive learning
- misclassification costs
- minority class
- cost sensitive
- class distribution
- multi class problems
- binary classifiers
- class imbalance
- total cost
- classification error
- naive bayes
- training examples
- multi class
- binary classification
- roc curve
- active learning
- class labels
- probability estimation
- random sampling
- missing values
- single class
- error rate
- generalization error
- small number
- support vector
- decision trees
- feature selection
- cross validation
- sample size
- multi class classification
- model selection
- rule extraction
- support vector machine
- sampling methods
- pairwise