Sign in

Dr. Rob Taylor

@dataforyou.bsky.social
15 followers 33 following 7 posts

R | statistics | data science | cognitive science | 🚲 🍻 🐕‍🦺 Research Officer @ Massey University

PostsRepliesMedia
Dr. Rob Taylor @dataforyou.bsky.social · 03/10/2026
Had some fun writing this one. Thanks to @towardsdatascience.com for publishing it! towardsdatascience.com/what-the-rel...
towardsdatascience.com
What the ReLU Revolution Revealed About Biological Plausibility
The activation function was never a fixed biological commitment but a working hypothesis revised under empirical pressure.
000
Reposted by Dr. Rob Taylor
Towards Data Science @towardsdatascience.com · 02/10/2026
"The pattern is a familiar one. The sigmoid gave way to ReLU because depth exposed its limitations, and ReLU gave way to GELU and its successors because scale and new architectures exposed its own." Rob Taylor shares an accessible history of activation functions.
towardsdatascience.com
What the ReLU Revolution Revealed About Biological Plausibility
The activation function was never a fixed biological commitment but a working hypothesis revised under empirical pressure.
011
Dr. Rob Taylor @dataforyou.bsky.social · 23/04/2026
Something I keep returning to when reflecting on my training: stats was taught as a checklist, not a way of thinking. Worse, the curriculum shaped what questions we thought to ask, not just how we answered them. New piece on what that costs, and what it could have been. medium.com/@dataforyou/...
medium.com
Statistical Technicians
The case for teaching statistics as reasoning, not procedure
000
Dr. Rob Taylor @dataforyou.bsky.social · 17/04/2026
Do we still need to test for equal variances before running ANOVA? Probably not as a decision rule. Welch's ANOVA, robust SEs, and variance-modelling approaches handle heterogeneity directly — assumption tests have shifted from gatekeepers to descriptive companions. medium.com/p/do-we-real...
medium.com
Do We Really Need to Test Variances Anymore?
Why classical assumption checks are no longer gatekeepers, and how modern ANOVA methods handle heterogeneity directly.
000
Dr. Rob Taylor @dataforyou.bsky.social · 11/04/2026
Revised post: Prussian soldiers, horse kicks, and Bayesian inference 🐎 I've revamped an old post on the Gamma-Poisson model — more historical context, a cleaner derivation, and a new section showing how the posterior updates sequentially across 20 years of data. medium.com/p/kicking-it...
medium.com
Death by Horse Kick: A Bayesian Tale from the Prussian Army
What a 19th-century military dataset can teach us about probability, belief, and uncertainty
030
Dr. Rob Taylor @dataforyou.bsky.social · 11/04/2026
Another one from the archives. This was actually the first blog I ever wrote. In it I cover what multicollinearity is and when it becomes an issue for modellers. Enjoy 😊 medium.com/data-science...
medium.com
Multicollinearity: Problem, or Not?
A brief guide on multicollinearity and how it affects multiple regression models
000
Dr. Rob Taylor @dataforyou.bsky.social · 11/04/2026
This is one from the archives. A while back a provided a discussion about the difference between probability and likelihood. Hopefully it’s still a useful piece for someone out there 🙂 medium.com/data-science...
medium.com
On Probability versus Likelihood
A discussion about two terms that are often used interchangeably, but are mathematically distinct
010
Dr. Rob Taylor @dataforyou.bsky.social · 10/04/2026
Hi Bluesky community 👋 I’m a data scientist and mathematical modeller with a background in cognitive science, where I used computational models to study how the brain represents information. Now I build data tools and write about making complex ideas clear and useful. Glad to be here!
161