Is a useful algorithm to find groups (clusters) in data. The algorithm let's you group a huge number of points (2d) into k clusters. Read more.
The algorithm first places k random points as centroids for the points and then fits them into the given data. Afterwards you do two steps in a loop:
- Find the nearest centroid for each point
- Iterate over all centroids and check if you should modify the position to be more in the center of the children.
- Set the flag to iterate one more time if you find one centroid whose position should be modified
Result looks like this:
You will have k groups of points that you can now further process.
