Hi community,
I was going through the Week 1 module in Clustering (Unsupervised Learning). My question is - would all points always fall in one cluster only?
In the T-shirt size example, wouldn’t some points be equidistant from centroid S and centroid M?
Hello @S1114,
There are examples in the lectures that show the points don’t always fall in one cluster only. Why do you think so? Which video are you talking about?
The example above? Even if there is, then we can just pick one of the centroids for that point.
Cheers,
Raymond
If they are equidistant from two centroids, then it does not matter which one they are assigned to.
That was the example I was talking about! Thanks!
Just a small clarification - in such cases, would the model assign the point to any one cluster (ensuring mutually exclusive clusters) OR will it give 2 assignments and we can choose?
The model would assign to one cluster - but this would be an exceedingly rare anomaly, as the centroid distances are floating point values. It is not worthy of very much consideration.
