
Select all of the following statements which are TRUE.

In order to identify dense regions, we need two hyperparameters:
eps: determines what it means for points to be “close”min_samples: determines the number of neighbouring points we require to consider in order for a point to be part of a cluster
Consider DBSCAN in a social context:

Experience how \(\epsilon\) (eps) and min_samples affect what counts as a cluster.
| Side of Class | \(\epsilon\) (eps) | min_samples | Expected Outcome |
|---|---|---|---|
| Left side | ~1 m (arm’s length) | 5 | Moderate clusters; Some isolated ones become noise points. |
| Right side | ~1 m (arm’s length) | 15 | Few or no clusters; the density requirement is high. Many noise points. |
eps and min_samples?GridSearchCV and RandomizedSearchCV?[ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15]

fclusterSuppose you want to go from 3 clusters to 2 clusters. Which clusters would you merge?
Select all of the following statements which are True
Discuss the following
| Clustering Method | KMeans | DBSCAN | Hierarchical Clustering |
|---|---|---|---|
| Approach | |||
| Hyperparameters | |||
| Shape of clusters | |||
| Handling noise | |||
| Distance metric |
Which clustering method would you use in each of the scenarios below? Why? How would you represent the data in each case?

