AntilliaAntillia
Contact ↗ PL
MachineSentinel / Recorded study

How well are anomalies detected?

Explore the recorded study results. Changing a selection shows another part of the report.

PR AUC

Higher values indicate better performance on this metric.

Autoencoder
0.2665
Isolation Forest
0.1563

ROC AUC

Higher values indicate better performance on this metric.

Autoencoder
0.7768
Isolation Forest
0.8654

Precision

Higher values indicate better performance on this metric.

Autoencoder
24.59%
Isolation Forest
13.77%

Recall

Higher values indicate better performance on this metric.

Autoencoder
29.41%
Isolation Forest
37.25%

Reading the result

The model misses many failures. This is a synthetic-data study, not an industrial safety system.