precisionRecallDataFrame
A data frame listing the precision and recall percentages for each class.
Declaration
var precisionRecallDataFrame: DataFrame { get }Discussion
Precision and recall are metrics calculated for each class. Together they describe the tradeoff between misapplying a label too liberally and missing examples of that label.
Precision describes how effective the model was at applying a label only when appropriate for a given category (few false positives).
Recall describes how effective the model was at finding all the relevant examples of a category (few false negatives).
[Image]
The figure below shows how each example contributes to the precision and recall percentages for the category “Elephant”.
[Image]
“Elephant” appears as the true or correct label only once, but it’s predicted twice. This second prediction is an error in precision. Precision and recall can give you a much better idea of how your model is making mistakes than classificationError.
To determine what other categories “Elephant” examples may have been labeled with, see the confusion property.