Name: AI Anomaly Detection Lab
Author: MySimulator

Anomaly score threshold

Threshold: 0.65

This case study models a vibration sensor on a factory conveyor motor. Each reading gets an isolation-style anomaly score based on how far it sits from the dense cluster of normal readings in a 2D feature space (vibration amplitude, temperature). Higher scores mean more isolated, more likely a genuine fault.

Outcome at this threshold

Faults caught—

False alarms—

Faults missed—

Sensor readings (red = flagged anomaly, gold ring = true fault)