Struct MLModelMetrics 

pub struct MLModelMetrics<A: Float + Send + Sync> {
    pub accuracy: A,
    pub precision: A,
    pub recall: A,
    pub f1_score: A,
    pub auc_roc: Option<A>,
    pub false_positive_rate: A,
    pub training_time: Duration,
    pub inference_time: Duration,
}

Performance metrics for ML models

Fields

accuracy: A

Accuracy of anomaly detection

precision: A

Precision (true positives / (true positives + false positives))

recall: A

Recall (true positives / (true positives + false negatives))

f1_score: A

F1 score

auc_roc: Option<A>

Area under the ROC curve, traced over every threshold the retained labelled scores admit, or None while fewer than two labelled scores of either class have been fed back.

This is deliberately optional. A detector’s confusion matrix alone fixes exactly one point on the ROC curve, so no area can be derived from it; the single-operating-point substitute (TPR + TNR) / 2 that used to be reported here was a different statistic wearing the AUC’s name. Call DetectionCounters::auc_roc for the reason it is unavailable.

false_positive_rate: A

False positive rate

training_time: Duration

Training time

inference_time: Duration

Inference time per sample

Trait Implementations

impl<A: Clone + Float + Send + Sync> Clone for MLModelMetrics<A>

fn clone(&self) -> MLModelMetrics<A>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<A: Debug + Float + Send + Sync> Debug for MLModelMetrics<A>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.