Statistician, ecologist, gull lover at UBC. Officially, I’m an Associate Professor of Measurement, Evaluation, and Research Methodology.
Vancouver, British Columbia
Joined November 2015
- Tweets1.4K
- Following1.1K
- Followers1.8K
- Likes113K
Pinned Tweet
The culmination of nearly a decade of work: my new paper with Louise Blight @procellaria11 and Wilson Cao on urban-nesting Glaucous-winged Gulls (GWGUs) of @CityofVancouver now online in Ecological Modelling. I'll give the highlights here, but more to come soon. 1/7
Ed Kroc retweeted
With almost any other valued skill, “ I’m terrible at it. I hate it” is either embarrassing or you'd be expected to explain. With math, people say it proudly across cultures and educational backgrounds
Sociologists should study this!
Ed Kroc retweeted
Remember we no longer have snow days, we have ZOOM days. A major winter storm is no reason to miss class!
Ed Kroc retweeted
Happy New Year everyone! Wishing you all the best for 2026 🥰 I’d love you to watch my 35 minute narrated video “Best of 2025” if you have time 😄 and also read my New Year blog post! I’d really appreciate your support! Thank you, Kate 🥰❤️🇨🇦
patreon.com/posts/147106514?…
I've often wondered how this result generalizes. (1) For which (prob) measures must their mass accumulate at the boundary of the ball as N blows up? (2) Given a (prob) measure, which shapes have this property? (Convexity seems important.) I'd love to know the answers.
This is one of the weirdest things I’ve ever learnt.🤨😦🤯
We’ve been so conditioned to think in 1, 2, and 3 dimensions that our intuition basically lives there, but once you step into high dimensions even something as basic as distance starts behaving in a way that feels wrong.
In a high-dimensional ball, almost all the volume lives in a thin shell near the boundary...shrink the radius just a tiny bit and you’ve thrown away almost everything, so random points don’t sit in the middle at all, they’re crushed into a microscopic halo at the edge.
That’s one of the reasons distance, nearest neighbours, and geometric intuition start acting so weird in high-dimensional Machine Learning spaces. High-dimensional geometry is quietly telling you your low-dimensional brain is lying to you. 🤯 #HighDimensionalSpace #MachineLearning
Lol, too real.
37 year old Jerry Seinfeld in his prime quarterbacks a masterpiece with some of the funniest people on the planet. David Spade, Adam Sandler, Chris Farley, Chris Rock & Siobhan Fallon deliver #comedyroyalty #snl 1992
Ed Kroc retweeted
"Wrong Way" Corrigan, the man who once flew all the way to Ireland by mistake, died this week at the age of eighty-eight.
"It was just like him," said his wife, "He was trying to live."
Blame the incentive structure. The more papers you publish and the more students you "train", the easier it is to get bigger grants, course buyouts, prestigious "Distinguished Professor of Blah"ships. There's no way to fix this.
Lol, this is all classical stats. My greatest criticism of statisticians is they have let their entire field be hijacked and rebranded by the AI/ML salesmen, and they don't seem to care. Hell, many (most?) go along with it.
Regularization techniques prevent machine learning models from overfitting—memorizing noise instead of learning patterns. Methods like L1 (Lasso) and L2 (Ridge) add a penalty to the loss function, shrinking weights to simplify the model. This ensures the AI generalizes well to new, unseen data. In real life, this robustness is critical for autonomous driving to ignore sensor noise and for medical AI to make reliable diagnoses across different hospitals.
Ed Kroc retweeted
The dean of the university asks the head of the physics department: "Why do you need such expensive equipment? Why can't you take a cue from the math department? All they ask for is pencils, paper, and wastebaskets! Or the philosophy department? All they ask for is pencils and paper!"
Ed Kroc retweeted
The collapse of the wavefunction doesn't break causality
Consider a lottery where many people buy tickets. Before drawing the winning numbers, the probability distribution of winners is a complicated function (depending on the number of participants, tickets per person,..)
1/7
Ed Kroc retweeted
Nonlocal correlations don’t break causality
The outcome of measurements can be correlated at space-like distances without contradiction even classically:
A bag has one red and one blue ball and we each take one, then even if you move across the galaxy your seeing red lets you infer mine is blue instantly. The measurement outcomes are correlated (at any distance) But no information has been exchanged faster than the speed of light
In Quantum Mechanics the correlation can be an entangled state (which doesn't exist classically):
In this case the color of the ball you're looking at is only determined at the time of measurement. The other ball is projected into eigenstate blue if you measure eigenstate red. But for each observer their individual measurement outcome is random. Again, no information has been exchanged faster than the speed of light
Bell was right, and the EPR paradox was misinterpreted by Einstein, because quantum mechanics is correct.
But, EPR is still a genuine paradox at the intersection of quantum mechanics and special relativity. We still lack a satisfying account of how nonlocal correlations can coexist with a relativistic notion of causality.
The no-signaling theorem prevents this apparent contradiction from corrupting our universe, but it’s still conceptually unresolved.
