Model Training & Adaptation
K-means
A method that sorts examples into a chosen number of groups around shared centers.
Example
A collection of information is divided into five groups based on selected numerical features.
Why people use it
It provides a straightforward way to divide numerical examples into a chosen number of groups.
What you'll hear
“Try grouping these cases around five centers.”
What this means for you
Check whether the chosen number and distance assumptions fit the data.
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 a very unusual example pull a group off course?
- Yes. Extreme values can shift a group's center and affect where other examples are assigned.
- Can different starting centers change the result?
- Yes. The method can settle on different groupings from different starting points.
- Does it suit every kind of group shape?
- No. Some shapes or very unequal group sizes fit its assumptions poorly.