More silliness

Back before I had so many followers, and it was less stressful to put goofy stuff “in the wild”, I wrote data science parody lyrics to “Summer of ’69” and “For the Love of Money”. Well, a while ago, another idea popped into my head..

and I was reminded of it today by this tweet

So, I wrote it down to share my silliness with you. Enjoy!

Lyrics:

I can fit every point
feature high info gain
but I’m over you asking me when you know I’m not okay
I learned really well
from the training that night
and though I can classify you know I’m just not right

I can’t generalize
I’m overfit
Too complex
I’m overfit
Tricked by noise
I’m overfit
I’m overfit
I’m over

Wanting you to be wanting me
But, my low accuracy
To outliers I’m sensitive
Because I’m so over
I’m overfit
Moving on, it’s too late
I didn’t pass cross validate
Tried so hard to relate
and now I’m so… so overfit!

You can go get more points
Build more models to help
You can regularlize with layers that drop out

Oh that’s why I’m done
I’m overfit
Was fun but
I’m overfit
I’m not the one
I’m overfit
I’m overfit
I’m over

Wanting you to be wanting me
But, my low accuracy
To outliers I’m sensitive
Because I’m so over
I’m overfit
Moving on, it’s too late
I didn’t pass cross validate
Tried so hard to data relate
and now I’m so… so overfit!

Don’t ask me to classify
Ain’t no use don’t ask me why
I’ll never change
Unless… tweak my parameters and run me again

Wanting you to be wanting me
But, my low accuracy
To outliers I’m sensitive
Because I’m so over
I’m overfit
Moving on, it’s too late
I didn’t pass cross validate
Tried so hard to relate
and now I’m so… so overfit!

(P.S. here’s the Wikipedia article on overfitting in machine learning!)