K-means clustering: place k centres, assign every point to its nearest centre, move each centre to the mean of its points, repeat until nothing changes. The animation runs the real algorithm on simulated data.
Disclosure: researched and scripted with Claude, an AI by Anthropic. Synthetic voice: Microsoft neural text-to-speech. Synthesised music. Animations are our own.
Sources:
– Lloyd (1982, written 1957), Least squares quantization in PCM:
– scikit-learn, K-means:
#MachineLearning #DataScience
Website:
LinkedIn:
GitHub:
Contact: akanda@akandaashraf.com
source

Notes: results depend on the starting centres, so implementations try several starts (k-means++). For soft, oval-shaped groups, see our Gaussian mixture video. Credit: Stuart Lloyd (1957), James MacQueen (named it, 1967).
Disclosure: researched and scripted with Claude, an AI by Anthropic. Synthetic voice: Microsoft neural text-to-speech. Synthesised music. Animations are our own.