Metrics

Includes classes for calculating the following:

  • Classification performance metrics

  • Unsupervised performance metrics

  • Pairwise metrics

Classification Metrics

Precision

Calculates the Precision for either a single or set of binary predictors.

Recall

Calculates the Recall for either a single or set of binary predictors.

FScore

Calculates the Fbeta score for either a single or set of binary predictors.

Revenue

Calculates the revenue for either a single or set of binary predictors.

Pairwise Metrics

CosineSimilarity

Computes the cosine similarity between columns in X.

JaccardSimilarity

Computes the Jaccard similarity between columns in X.

Unsupervised Metrics

AlertsPerDay

Calculates the negative squared difference between the number of alerts per day in the binary predictor(s) vs the expected.

PercVolume

Calculates the negative squared difference(s) between the percentage of the overall volume that the binary predictor(s) vs the expected.