Metrics used in information retrieval and pattern recognitions are going to be more concerned about identifying positives, i.e. identifying a single class.
- Precision: Fraction of retrieved instances that are relavant
- Also known as positive predictive value
- (# of predicted positive outcomes/Total # of positive outcomes)
- [Recall] (http://stats.stackexchange.com/questions/8089/why-doesnt-recall-take-into-account-true-negatives): Primarily used
as a metric in information retrieval. Refers to the fraction of relevant instances that were retrieved.
- AKA sensitivity
- Type I error:
- Type II error;
- Accuracy
- RMSE
- MSE
- MAE