The Elbow Method Explained in Less than 5 minutes

[@sds-superdatascience]

Thank you for subscribing and following our videos. You can find the course HERE:sds.courses/python-ml-level-1

[@zabairbhatti7548]

Exceptionally well explained! Thank you very much!

[@Carlitos_SH]

Simply explained, Thanks !

[@markvogt70]

INFORMATIVE & ENJOYABLE video - you've earned yet another subscriber ! ONE COMMENT (a correction you may wish to make at your earliest opportunity, since this has been out a year): GIVEN your intro set of 30 (THIRTY) data points, your "Elbow Method" Graph (timestamp 3:26) couldn't possibly end up with WCSS = 0 at only 10 (TEN) clusters :-O INSTEAD the "Zero WCSS Point" on the x-axis would be for precisely 30 clusters, each with 1 data point which by definition of WCSS has a value of 0 (ZERO). You explain this VERBALLY quite nicely, but I think your GRAPH then contradicts what you've explained, because at Clusters = 10 there would be approximately 3 data points in each cluster (on average) hence a NON-ZERO WCSS for each cluster. I look forward to watching more of your videos - you're an EXCELLENT presenter !! - Mark Vogt, Principal Solution Architect/Data Scientist - Avanade (avanade.com)

[@neginghaheri3285]

It was great , thanks !

[@klarwieglas8144]

Hey, Do you have a source for the video where the WCCS is explained again? Great video by the way

[@brendansmith5529]

I don’t understand how a visual estimation of the “elbow” tells us anything mathematically. Assuming something like an exponential asymptotic approach to a “best” value, wouldn’t this elbow point apparently change with different x axis scales? Or maybe some function families have an “elbow” and others don’t. Or could it maybe be something like an inflection point is for continuous functions? Very interesting!

[@LeHuuHuy_]

i don't understand why elbow method is optimal for dbscan and k-mean ? Someone can explain this. Thank you!

[@NilsThomas19]

Simply explained, Thanks !