Functional Metrics¶
Aggregation Metrics¶
Computes Area Under the Curve (AUC) using the trapezoidal rule. 

Compute weighted mean. 

Compute weighted sum. 

Calculate the throughput value which is the number of elements processed per second. 
Classification Metrics¶
Compute binary accuracy score, which is the frequency of input matching target. 

Compute AUPRC, also called Average Precision, which is the area under the PrecisionRecall Curve, for binary classification. 

Compute AUROC, which is the area under the ROC Curve, for binary classification. 

Compute AUROC, which is the area under the ROC Curve, for binary classification. 

Compute precision recall curve with given thresholds. 

Compute binary confusion matrix, a 2 by 2 tensor with counts ( (true positive, false negative) , (false positive, true negative) ) 

Compute binary f1 score, the harmonic mean of precision and recall. 

Compute the normalized binary cross entropy between predicted input and groundtruth binary target. 

Compute precision score for binary classification class, which is calculated as the ratio between the number of true positives (TP) and the total number of predicted positives (TP + FP). 

Returns precisionrecall pairs and their corresponding thresholds for binary classification tasks. 

Compute recall score for binary classification class, which is calculated as the ratio between the number of true positives (TP) and the total number of actual positives (TP + FN). 

Returns the highest possible recall value given the minimum precision for binary classification tasks. 

Compute accuracy score, which is the frequency of input matching target. 

Compute AUPRC, also called Average Precision, which is the area under the PrecisionRecall Curve, for multiclass classification. 

Compute AUROC, which is the area under the ROC Curve, for multiclass classification. 

Compute AUROC, which is the area under the ROC Curve, for multiclass classification. 

Compute precision recall curve with given thresholds. 

Compute multiclass confusion matrix, a matrix of dimension num_classes x num_classes where each element at position (i,j) is the number of examples with true class i that were predicted to be class j. 

Compute f1 score, which is defined as the harmonic mean of precision and recall. 

Compute precision score, which is the ratio of the true positives (TP) and the total number of points classified as positives (TP + FP). 

Returns precisionrecall pairs and their corresponding thresholds for multiclass classification tasks. 

Compute recall score, which is calculated as the ratio between the number of true positives (TP) and the total number of actual positives (TP + FN). 

Compute multilabel accuracy score, which is the frequency of input matching target. 

Compute AUPRC, also called Average Precision, which is the area under the PrecisionRecall Curve, for multilabel classification. 

Returns precisionrecall pairs and their corresponding thresholds for multilabel classification tasks. 

Returns the highest possible recall value give the minimum precision for each label and their corresponding thresholds for multilabel classification tasks. 

Compute multilabel accuracy score, which is the frequency of the top k label predicted matching target. 
Ranking Metrics¶
Compute the click through rate given a click events. 

Calculate the frequency given a list of frequencies and threshold k. 

Compute the hit rate of the correct class among the top predicted classes. 

Compute the number of collisions given a list of input(ids). 

Compute the reciprocal rank of the correct class among the top predicted classes. 

Compute weighted calibration metric. 
Regression Metrics¶
Compute Mean Squared Error, which is the mean of squared error of input and target Its class version is 

Compute Rsquared score, which is the proportion of variance in the dependent variable that can be explained by the independent variable. 
Text Metrics¶
Compute BLEU score given translations and references for each translation. 

Perplexity measures how well a model predicts sample data. 

Compute the word error rate of the predicted word sequence(s) with the reference word sequence(s). 

Compute the word information preserved score of the predicted word sequence(s) against the reference word sequence(s). 

Word Information Lost rate is a metric of the performance of an automatic speech recognition system. 