The Very Model of a Professor Statistical

The last chapter of George Box’s book Improving Almost Anything contains the lyrics to “I Am the Very Model of a Professor Statistical,” to be sung to the tune of “I Am the Very Model of a Modern Major General” by Gilbert & Sullivan.

Here’s the original:

The original song has a few funny math-related lines.

I’m very well acquainted, too, with matters mathematical,
I understand equations, both the simple and quadratical,
About binomial theorem I’m teeming with a lot o’ news,
With many cheerful facts about the square of the hypotenuse.

I’m very good at integral and differential calculus;
I know the scientific names of beings animalculous:
In short, in matters vegetable, animal, and mineral,
I am the very model of a modern Major-General.

Here are a few lines from George Box’s version.

I relentlessly uncover any aberrant contingency
I strangle it with rigor and stifle it with stringency
I understand the different symbols be they Roman, Greek, or cuneiform
And every distribution from the Cauchy to the uniform.

With derivation rigorous each lemma I can justify
My every estimator I am careful to robustify
In short in matters logical, mathematical, idealistical
I am the very model of a professor statistical.

Gilbert & Sullivan have come up on this blog a couple other times:

George Box has come up too, but only once. (I’m surprised he hasn’t come up more; I should rectify that.) This post has a great quote from Box: “To find out what happens to a system when you interfere with it, you have to interfere with it (and not just passively observe it).”

5 thoughts on “The Very Model of a Professor Statistical

  1. A treasure. Thanks for sharing.
    The importance of history is captured in

    “understand the different symbols be they Roman, Greek, or cuneiform”

    There is also taste in the choice of topics:

    “And every distribution from the Cauchy to the uniform”

    these are exactly the ones which are invariant when adding
    independent random variables.

  2. I first encountered George Box via the Box-Muller Transform (BMT) in the mid-1990s, when I needed to simulate the signal from a radiation detector. The simulation was needed to test the reactor power (neutron) channel of a reactor control system, most critically the SCRAM triggers, to ensure proper triggering (with minimal false positives and NO false negatives) in real-time using an underpowered microprocessor. The combination of the uniform POSIX random() source with the BMT seemed to provide all the “knobs I needed to twist” to meet my simulation needs.

    The first BMT implementation I encountered seemed deceptively simple. Before relying on it for a critical application, I dove down the derivation and proof rabbit hole. Which fortunately yielded insights that in turn yielded an optimized version that, after exhaustive testing, worked very well for my application domain and target platform.

    The same code was then used as part of the verification testing for all neutron channel instrument code releases, as it also provided the ability to precisely mimic reactor power transients that had previously caused instrumentation errors and unneeded SCRAMs. Which was the original problem I was brought in to investigate and (hopefully) resolve.

Comments are closed.