Songs: Random Stratified Sampling

notes / gigs / releases

Because if we've got a corpus
Of two hundred and sixty six
To be frank I haven't got time
To analyse all of this

So what we need is a sample
About a quarter the size
But with the same proportions in terms of
Series title and time

It needs to be representative
And the statistical technique I used to do it with is

Random Stratified Sampling

The choices were made at random
And empirically
So that the final selection
Had nothing to do with me

If it had I assure you
That the sample size
Would not have worked out to be
69

But still it means that when I describe
My research people always say it’s "Nice!"

With Random Stratifed Sampling

Published by Wipe Out Music Publishing

This may be my favourite song in the whole of Data and Doctor Doom because, of all of them, it feels the most like a Validators song. The lyrics are not AT ALL like that, but playing it feels like it should have them all joining in!

The song is all about the methodology I used to generate a sample for my research, and the bit in the middle is where I stop singing and then EXPLAIN the whole thing to an audience with an example of some SHAPES. I originally did this demo as part of a presentaton I did with Dr Ian Horton at IGNCC in 2024 called Digital Databases and Donald Duck Weekblad (or Double Data Entry is Double Dutch to Me) (shockingly neither the pun NOR the amazing alliteration were my idea for this one - I wish they were!), and I liked it so much I decided to keep it for this.

The only slight SHADOW cast over this one is that, once it was all finished, I went and re-read some info about the methodology and realised it was actually called "RandomISED Stratified Sampling." I did have a go at changing it, but pretty quickly decided to exercise POETIC LICENCE and leave it as it. ROCK AND ROLL!