k-means clustering - explained

See all my videos at www.tilestats....
1. How the method works
2. How to calculate the within-cluster sum of squares (05:37)
3. How to select k based on the elbow method (08:15)

[@merveak1129]

You are my hero .My major degree is statistics and I believe the theory to understanding these subjects than coding .You're making easy to understanding in complex things in book .I'm so appricate to share your knowledge to us .

[@user-wr4yl7tx3w]

best explanation on K Means. I'm not sure why others can't explain it as well when the concept is relatively straightforward, especially with regard to WCSS. I came across multiple medium articles where it made WCSS more confusing.

[@daytodatainc.1262]

One of the best videos I’ve watched ok K-means and the best explanation of how to use it! Thank you, this really helped me understand the use in conjunction with data. Also a proper explanation of the methods to be used to determine # of centroids needed using the Elbow method. 💪💪💪

[@casper8374]

underrated channel, hidden gem

[@ehsanakbari3519]

that was great , Thank you for your great explanation

[@aryankashyap7194]

00:03 K-means clustering divides data into k clusters. 01:19 Determining the value of k is essential for k-means clustering 02:41 k-means clustering assigns data points to the nearest centroid and updates centroid positions 04:11 Clusters shifting based on proximity to centroids 05:36 Measuring cluster performance with within-cluster sum of squares 06:58 Choosing the best clustering output based on within cluster sum of squares. 08:22 Optimal value of k in k-means clustering 09:39 The optimal number of clusters for the example data is 3.

[@pramitthapa283]

However, he already started with k=3

[@casper8374]

I have a question, which one should I do first ? elbow method to find the k, or try different starting random centroids for a certain k ?

[@tilestats]

Elbow last once you have find the best starting pos for the ks.

[@nelsonk1341]

thanks ! keep up the great work!

[@tilestats]

Thank you!

[@jano979797]

Great work. One Querstion. How do i calculate the midpoint of the dataset?

[@tilestats]

Have a look at this video where I calculate the centroid kzbin.info/www/bejne/ronLfamemqp5bdE

[@user-ns8rn8fu3z]

Hi sir is k means and kneighborhood algorithms are same ?

[@tilestats]

No, have a look at this video for KNN kzbin.info/www/bejne/amm1ootqfbmneac

[@pramitthapa283]

Like other youtube teachers, failed to explain why k=3 was chosen, why not 2 or four.

[@tilestats]

It is explained at 8:12 and forward.

[@pramitthapa283]

@@tilestats Thanks. I see now. I was expecting earlier, and stopped watching (without understanding chosen value of K) the complete video

[@codework-vb6er]

@TileStats @7:00 what are your values for xhat and Xhat? my BCSS = 2.0 * 7.817901234567899 + 6.0 * 26.382716049382715 + 2.0 * 9.373456790123456 = 192.679012345679 I used as my Xhat the mean of all xhat's, which is [8.5, 7.888888888888889]. my set of xhat's are [9.5, 10.5], [4.5, 4.666666666666667] [11.5, 8.5]. Great Tutorial!

[@tilestats]

Xhat should be the mean of all data points, or the weighted mean of the three xhats.