Module org.elasticsearch.xcore
Package org.elasticsearch.xpack.core.ml.dataframe.evaluation.classification
package org.elasticsearch.xpack.core.ml.dataframe.evaluation.classification
-
ClassesClassDescription
Accuracyis a metric that answers the following two questions: 1.Area under the curve (AUC) of the receiver operating characteristic (ROC).Evaluation of classification results.MulticlassConfusionMatrixis a metric that answers the question: "How many documents belonging to class X were classified as Y by the classifier?" for all the possible class pairs {X, Y}.Precisionis a metric that answers the question: "What fraction of documents classified as X actually belongs to X?" for any given class X equation: precision(X) = TP(X) / (TP(X) + FP(X)) where: TP(X) - number of true positives wrt X FP(X) - number of false positives wrt XRecallis a metric that answers the question: "What fraction of documents belonging to X have been predicted as X by the classifier?" for any given class X equation: recall(X) = TP(X) / (TP(X) + FN(X)) where: TP(X) - number of true positives wrt X FN(X) - number of false negatives wrt X