Measurement bias

Measurement bias occurs when the data collected is faulty measured and when it differs from the real-world data.

Example: When you record the audio to train an AI model with one microphone and then use another microphone during the production.

Example: When a quality inspector defines the defective products as non-defective and uses that data to train a model.

Defective and non defective products

How to prevent measurement bias

To prevent measurement bias, you can check:

  • If the measurement tool is working properly

  • If there is an inconsistency between measurements with several tools

  • If the data collected is faulty by human mistake

Was this page helpful?
Yes No