Module 1 -> Clustering -> k-means optimization objective

[vinooj]

When creating a post, please add:

• Week # must be added in the tags option of the post.
• Link to the classroom item you are referring to:
• Description (include relevant info but please do not post solution code or your entire notebook):

1. In the attached slide, if you have only m clusters, shouldn’t we only have m cluester centroids as well?
   
   

What did I miss?

[TMosh]

In the slide, the number of clusters is ‘K’.
‘m’ is the number of examples.

There may be an error in the slide.

[Hari_Krishnan_94]

@vinooj

have only m clusters, shouldn’t we only have m cluester centroids as well?

Your question is not clear to me. I’ll still try to answer as far as possible. m represents the total number of data points we have and for each of such data points there’ll be cluster index (so indirectly a cluster centroid) assigned to them.

We compute the error from each data points, as we did for the supervised learning, to obtain the cumulative error. And Error from each data here is defined as the L2 norm distance of each data point to its corresponding cluster centroid.
Hope it helps!
-Hari

[rmwkwok]

Hello @vinooj, could you be more specific about which part of this slide has a problem? The slide didn’t say there were m clusters, instead it said there were K clusters.

[vinooj]

I was getting little confused about the J(c(1)…c(m),µ1…µk). I was under the impression if there are m cluster indices, then there will be µm centroids as well.

[vinooj]

Thanks for helping

[vinooj]

See my comments

[rmwkwok]

You are welcome, @vinooj.