Package jsat.classifiers.evaluation

Interface Summary

Interface	Description
ClassificationScore	This interface defines the contract for evaluating or "scoring" the results on a classification problem.

Class Summary

Class	Description
Accuracy	Evaluates a classifier based on its accuracy in predicting the correct class.
AUC	Computes the Area Under the ROC Curve as an evaluation of classification scores.
F1Score	The F1 score is the harmonic mean of Precision and Recall.
FbetaScore	The Fβ score is the generalization of F1Score, where β indicates the level of preference for precision over recall.
Kappa	Evaluates a classifier based on the Kappa statistic.
LogLoss	This computes the multi-class Log Loss - 1/N Σ∀ i ∈ N log(pi, y) Where N is the number of data points and pi, y is the estimated probability of the true class label.
MatthewsCorrelationCoefficient	Evaluates a classifier based on Mathews Correlation Coefficient
Precision	Evaluates a classifier based on the Precision, where the class of index 0 is considered the positive class.
Recall	Evaluates a classifier based on the Recall rate, where the class of index 0 is considered the positive class.
SimpleBinaryClassMetric	This is a base class for scores that can be computed from simple counts of the true positives, true negatives, false positives, and false negatives.