What you’ll learn
Identify clusters and outliers.
Key idea
A scatter outlier is separated from the displayed pattern; it need not have the largest output.
Isolation concerns the whole displayed pattern, not maximum y; no formal distribution cutoff or automatic deletion.
Worked example
Classify examples and non-examples for the concept 'Identify clusters and outliers'.
Classify O relative to this observed point cloud, with separation evidence.
Marked coordinates: *=(-2, 2); *=(-1, 2); *=(-2, 1); *=(-1, 1); O=(7, -8).
- The cluster occupies a unit square: {(-2,1),(-1,1),(-2,2),(-1,2)}.
- The isolated pair is (7, -8); horizontal separation from the nearest cluster edge is 8, far larger than the cluster width.
Answer: O is an outlier — four observations cluster in x=-2..-1,y=1..2, but (7,-8) is isolated.
Quick Check
Classify examples and non-examples for the concept 'Identify clusters and outliers'.
Classify O relative to this observed point cloud, with separation evidence.
Marked coordinates: *=(0, -1); *=(1, -1); *=(0, -2); *=(1, -2); O=(6, -8).