Model Training & Adaptation
Feature scaling
Adjusting the number ranges of measurements so an AI can work with them appropriately.
Example
A team adjusts the ranges of income and age numbers before teaching AI.
Why people use it
It keeps different measurement sizes from distorting how some AI methods learn.
What you'll hear
“Put these measurements on comparable scales.”
What this means for you
Apply the same preparation rule consistently to later information.
Can you control it?
Developer-only
The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.
Common questions
- Can scaling rules learned from future records make a test unfair?
- Yes. That gives the preparation process information unavailable at the time a real prediction would be made.
- Does scaling change the basic facts?
- No. It changes their numerical description, much like expressing a length in different units.
- Does every method need it equally?
- No. Some methods are much more sensitive to measurement scale than others.