George Udny Yule

George Udny Yule (1871–1951) was a genteel, quietly witty Scottish statistician who asked a wonderfully strange question: why does the number of sunspots rise and fall in a rough, wobbly rhythm that never quite keeps time? His answer, in 1927, gave the world autoregression — the idea that a series can be driven by its own past. It is the beating heart of modern time series analysis, and it started with the Sun.

The pendulum and the boys with peas

Before Yule, people fitting curves to sunspots assumed a hidden clockwork sine wave buried under some measurement noise. Yule thought that was wrong. He offered instead one of the most charming metaphors in all of statistics: imagine a swinging pendulum. A tidy sine wave is a pendulum swinging in a silent room. But real data, he said, is a pendulum being pelted by boys throwing peas at it. Each little knock nudges it off course, so the swing keeps going but its timing and size wander — exactly like the sunspots. The randomness isn't sitting on top of the motion; it is feeding into the motion itself. That single shift of viewpoint is autoregression.

Equations that still bear his name

To fit these models Yule worked out the relationships now called the Yule–Walker equations (Gilbert Walker later extended them while studying the Indian monsoon). Give them the autocorrelations of a series and they hand you back the autoregressive coefficients — the recipe for the pendulum. A century on, they are still the first thing a student learns for estimating an AR model.

Yule also gave us a warning that every data scientist should tattoo somewhere visible. He coined the term "spurious correlation" (he liked to call them "nonsense correlations") after noticing that two utterly unrelated trending series — say, the number of storks and the number of babies, or the marriage rate and the price of imported apples — can look tightly linked purely because both drift over time. The correlation is real in the arithmetic and meaningless in the world. He proved it wasn't a fluke but a trap that time series lay for the unwary, and much of the machinery of the subject exists to avoid falling into it.

The textbook and the man

Yule learned his trade as an assistant to the formidable Karl Pearson at University College London, then spent most of his career at Cambridge. He wrote An Introduction to the Theory of Statistics, a book so clear and popular it ran to fourteen editions and taught the subject to a generation. Colleagues remembered him as courteous, funny and faintly eccentric — in later life he took up flying and bought his own aeroplane. A gentle man who, by watching the Sun's freckles, quietly rewired how we think about time.

Read more

The fuller story is on Wikipedia: Udny Yule — Wikipedia.