DBSCAN: Part 1

In this session, we are going to introduce a density-based clustering algorithm called DBSCAN. DBSCAN is a density-based spatial clustering algorithm introduced by Martin Ester, Hanz-Peter Kriegel's group in KDD 1996. This paper received the highest impact paper award in the conference of KDD of 2014. This paper developed an interesting algorithms that can discover clusters of arbitrary shape. Actually, DBSCAN itself is acronym of density-based spatial clustering of applications with noise.

[@hARRYnhariprasathnallasamy]

neatly explained sir! thanks

[@c.v.shankarshankar8774]

Thanks sir. It's quite simple but very effective explanation.

[@rodguinea]

This video is great! Thanks

[@billwindsor4224]

Hello Sir - excellent explanation of DBSCAN you have presented here, including computer science concepts of computational complexity, *thank you*. I am really looking forward to your next lecture on DBSCAN parameter control for Epsilon and MinPoints; because that is an important area for DBSCAN performance and effectivenesss, as you note (and there is relatively little information available on this!).

[@PandemicGameplay]

Nice explanation

[@psychorameses]

Based on this explanation, what's the difference between density-reachable and density-connected? Density-connected is defined as a point o such that both p and q are density-reachable from o, which means there exists two chains of points p1, p2, ..., o and o, ..., q2, q1 that are both fully directly density-reachable, which means p1, p2, ..., o, ..., q2, q1 is also a fully directly density reachable chain. Wouldn't this mean density-connected is essentially the same as density-reachable? Something's off here.

[@MachineLearningTV]

Dear friend, the definition of the point type in DBSCAN is a little confusing. Search in Google and you Will find good texts that complement this video. I also suggest you to watch part 2 of this video.

[@pred9990]

much more understandable and structured than indian guys explanation)

[@james-ht5vf]

4:46