DBSCAN

Question 1

What the density based do ?

Answer 1

locates cluster even if they do not belong regions of high density, and in any
separates outliers

Question 2

What is the two parameters in DBScan

Answer 2

MinPts

𝜀(epsilon):

Question 3

What is 𝜀(epsilon) ?

Answer 3

𝜀(epsilon): Maximum radius of the neighborhood for some point p➢ Neighborhood of p: points within a radius of from a point p

Question 4

What is minpts?

Answer 4

Minimum number of points in an neighborhood of a point

Question 5

Give me the Characterization of points

Answer 5

Core point
Border point
Noise point

Question 6

What the diff between core and border ?

Answer 6

Core has more than specific number of minPts

Border has fewer than minPts but has core in region

Question 7

Direct density reachable

Answer 7

The link between core and any point in region of core point

Question 8

Density-reachable

Answer 8

The link between core to another point in diffrent cluster but most be connect to other core points to access to the point

Question 9

Density-connected

Answer 9

The link between two point in different cluster but most be between them core point

Question 10

Why the epsilon important?

Answer 10

If is very small their will not core point

If very large the results in all point forming single cluster

Question 11

How determining EPS epsilon?

Answer 11

Support have minPts=4 then the EPS is the distance of 4 neighbour

Question 12

The pors of DBScan

Answer 12

Does not require the number of clusters a priori.
▪ it can capture arbitrary shape cluster
▪ it can identify (noises)

Question 13

Cons of DBScan

Answer 13

1-Fails in case of varying density clusters.
2- Does not work well of high dimensional data.
3-Sensitive to Parameters.