How does the 3d clustering work?

I was wondering how the 3d clustering decides which clusters it will create? In the example on plotly website the clusters are not defined anywhere in the data but somehow three clusters are made.

Can you link to the example?

@Jchaykow, Depending on your data, and the purpose of their clustering you can use one of the clustering methods, popular in machine learning: k-means, DBSCAN, t-sne or umap.