(There Must Be) 50 Ways to Cluster Data
[with apologies to Paul Simon]

The problem's all inside your head, she said to me,
Clustering is easy if you take it hierarchically.
Howard-Harris can also set you free,
There must be 50 ways to cluster data.

She said it's really not my habit to intrude,
But solutions are so numerous, both elegant and crude.
So I'll repeat myself, and I'm happy to collude,
There must be 50 ways to cluster data.

Try Univac, Jack, use a new Cray, Ray,
There's Honeywell, Mel, just listen to me.
Use Apple II Plus, Gus, no need to regress much,
Try IBM, Lem, and CDC.
Use MDS, Cal, or SPSS, Les,
There's Bio-Med, Fred, that's BMD-P.
Segment on race, Ace, cluster on sex, Rex,
Go hierarchically, and draw a nice tree.

She said there really is no basis for mystique,
You can rotate your axes, either varimax or oblique.
Get our your manuals and take a little peek,
There must be 50 ways to cluster data.

She said why don't we both just iterate tonight,
And I'm sure that by the morning we'll reach some new insight.
Then she KYST me and I realized she probably was right,
There must be 50 ways to cluster data.
--Paul Riedesel, c1983