Section 17 of 24 · 4 lessons

Anomaly Detection

An unexpected reading can mean a broken sensor, a real event, or an operating condition we have not seen before. We will compare several ways to recognise unusual records and turn their scores into an investigation rule. A high score will remain evidence to inspect, rather than a diagnosis.

0 of 4 finished

Recognising Unusual Records

An unusual reading needs a reference. Compare a robust global baseline with a local neighbourhood, and keep unusualness separate from a confirmed fault.

  1. 01Finding Unusual Readings Without Letting Them Set the ReferenceA faulty sensor can pull an ordinary mean toward itself. Use a robust centre and spread, then decide what a large deviation actually warrants.
    Build with Botie →Read the lesson
  2. 02Judging a Reading Against Its NeighboursDifferent operating regimes can have different spreads. Local outlier factor compares a point’s density with the density around its neighbours.
    Build with Botie →Read the lesson

Different Ways to Detect Unusualness

Some observations are easy to isolate; others are hard to reconstruct. Work through both signals before deciding which cases deserve investigation.

  1. 03Finding Readings That Are Easy to SeparateAn isolated reading often needs few random splits to stand apart. Follow paths through small trees before averaging that evidence across a forest.
    Build with Botie →Read the lesson
  2. 04Detecting Patterns a Model Cannot ReconstructA model trained to reproduce normal sensor patterns can expose departures through reconstruction error. The error still needs a validated alert rule.
    Build with Botie →Read the lesson