Re:Introduction to Mean Shift Algorithm
by Wookai on Friday September 02, 2016 @01:55PM
(#12414847)
Based on this plot, my naive reaction would be to decide that DBSCAN is the best algorithm, as it recovers the (qualitatively) best clusters in each case. Anybody has experience with DBSCAN? What are the downside?
[ Parent
| on HN ]
The Wiki page has a nice writeup of a few downsides: https://en.wikipedia.org/wiki/DBSCAN#Disadvantages