id in the kmeans function

[id,C,SUMD]= kmeans(Dataset,3);

[3:43 PM]

I understand the C refers to the centroid, SumD which is set to the squared euclidean distance by default, but I am unsure as to what id does? I tried finding the documentation where it says idx (which I assume is the same as id) gives the cluster indices, so does that mean that it numerates the clusters somehow?

Answer 1

[Image Analyst]

id is the class identification number that kmeans() thinks each of your data rows best belongs to. So it you had, say 1000 points and told it to assume 3 clusters (classes), then kmeans will classify each of your data rows into one of the 3 clusters (classes) and return id, which will be a column vector of 1000 rows where each value is either 1, 2, or 3 depending on what cluster/class number it thinks that particular data point most likely belongs to.