banner
LightNews
dperrin.bsky.social
Dimitri Perrin
@dperrin.bsky.social
150 followers 85 following 62 posts
Co-Director, QUT Centre for Data Science. Head of Data Science, Queensland Reds. Data Science, Machine Learning and AI applied to health, biology, and sports. 🇫🇷 -> 🇮🇪 -> 🇯🇵 -> 🇮🇪 -> 🇯🇵 -> 🇦🇺
Posts Media Videos Starter Packs
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 1d
If you are coming to IDW, you will see many of our Centre members Monday-Thursday, but you can also join us on Friday for further networking: events.humanitix.com/screens-stor...
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 2d
Congrats! 🥳
Reposted by Dimitri Perrin
robp.bsky.social
Rob Patro @robp.bsky.social · 3d
And it's posted! If you're interested and eligible, please consider applying through the UMD portal: umd.wd1.myworkdayjobs.com/en-US/UMCP/j....

If you're a PI working in algorithmic genomics (& you can recommend my lab to your top graduating students ;P), please let them know!
18 18
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 12d
It depends how we define fairness. Does a fair draw give teams an equal chance of reaching the round of 16 (irrespective of their strength), or does it give them a probability that is as close as possible to their strength? And do we maximise average fairness, or minimise the largest unfairness?
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 12d
Yes, I’ve noticed that it is what you have done, but I wasn’t sure what made it fair. It makes being ranked 6th the best possible ranking (assuming that the rankings actually reflect team strength).
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 13d
It depends what you mean by fair. Is it about giving relatively weaker teams a leg up, to increase competition, or is it about rewarding higher ranked teams?
1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 27d
You’re very welcome. Pre-ordered and all set to dive into it once it drops.
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 27d
Australia dropped to 7th, but looking at a single week does not give you the whole picture. You need to look at all possible results, so it is time to update the rankings simulations. Everything is still to play for, not just for the TRC. open.substack.com/pub/dimitrip...
Back to square one?
If everyone is special, no one is.
open.substack.com
1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · 29d
This should be a good read. I’m looking forward to it.
samlstandsup.bsky.social
Sam Larner @samlstandsup.bsky.social · Sep 11
That big news I teased yesterday?

I've written a book!

Attacking the Space: Inside Rugby's Tactical and Data Revolution is the book I wanted to read as a rugby obsessive kid. It will be released on the 29th January 2026, the pre-order link is:

geni.us/AttackingThe...

1)
1 1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 17
It might have been Jonathan Drennan’s SMH article. It was pretty good, but missing the impact of the sequence of results as well as the results from the other teams. That’s where the simulation tool becomes handy. It captures all games, rather than just the Wallabies.
1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 17
On one of your last points, on the impact on the rankings and a top-6 seeding for the World Cup, I have updated the little tool I wrote last week, and it is still far from certain. Even beating Argentina twice would not guarantee it: bsky.app/profile/dper...
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 17
I have updated my simulations to include the two results from the weekend. Great win for the Wallabies, but consistency will be key to get a top-6 seeding. It is still far from certain: open.substack.com/pub/dimitrip...
A win for the ages, but it does not guarantee top 6
Still a long road ahead before Australia gets a top seed for the World Cup draw
open.substack.com
1 1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 17
I have updated my simulations to include the two results from the weekend. Great win for the Wallabies, but consistency will be key to get a top-6 seeding. It is still far from certain: open.substack.com/pub/dimitrip...
A win for the ages, but it does not guarantee top 6
Still a long road ahead before Australia gets a top seed for the World Cup draw
open.substack.com
1 3 4
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 11
Thanks!
a man in an orange shirt says " thank you " to another man
ALT: a man in an orange shirt says " thank you " to another man
media.tenor.com
1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 10
Thanks! ☺️
1 1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 10
First post on substack. I’ve had a look at World Rugby rankings, and what they might look like by the time they really matter (when the RWC draw is made). It’s there: open.substack.com/pub/dimitrip...
The race for ranking points and a good RWC draw
A simulation-based look into the next few months of international rugby
open.substack.com
1 5 15
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Aug 7
Congrats! 🎉
1 1
Reposted by Dimitri Perrin
anshulkundaje.bsky.social
Anshul Kundaje @anshulkundaje.bsky.social · Jun 19
@saramostafavi.bsky.social (@Genentech) & I (@Stanford) r excited to announce co-advised postdoc positions for candidates with deep expertise in ML for bio (especially sequence to function models, causal perturbational models & single cell models). See details below. Pls RT 1/
1 40 55
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · May 20
I’ve also been tracking selections and minutes (but don’t have it on my phone). I’ll check later.
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · May 20
I’ll probably run out of time before the Super Rugby season ends, but I might try something later. Worst case, it’s ready for next season.
1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · May 20
There may be a way to do it in a way that is public friendly. I had something related for baseball, but without the idea of visualising it as the ladder itself. (though perhaps not the best example in terms of ease of interpretation; it can be improved)
1 1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · May 20
It takes a bit of scrolling, but the Super Rugby website has a list of injured players for each round.

An alternative could be to look at stability of the 23 week to week, as rotation can be enforced (injuries) or voluntary (Leinster, or Brumbies round 1 on a smaller scale).
1
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · May 20
Very nice indeed! For extra points, lighter background and the horizontal bar is a colour gradient based on the probability of each position for that team. 😁
1 2
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · May 20
Nice write up. Your last point on cohesion from 2024 into 2025 is important. I’ve looked at the impact for the last RWC, and it was quite striking (that will hopefully be out soon; writing in progress). For SRP, the impact of injuries on the cohesion week to week (or lack thereof) further adds to it
1 1
Reposted by Dimitri Perrin
sydonahi.bsky.social
Dr. Heloise Stevance @sydonahi.bsky.social · Apr 27
I can't emphasize how important this is. If you are not an expert or able to check AI outputs you cannot use them safely.

To add to this conversation here is recent work from Anthropic showing that REASONING models are NOT FAITHFUL

mini 🧵

(🔭🧪)

www.anthropic.com/research/rea...
9 48 130
dperrin.bsky.social
Dimitri Perrin @dperrin.bsky.social · Apr 11
It was a pleasure to talk to @paulcully.bsky.social a few days ago about my analysis of the Super Rugby Pacific season. His article is now up on the Sydney Morning Herald website: www.smh.com.au/sport/the-nu...
The numbers don’t lie – why Super Rugby is the best club comp in the world
A rugby-loving French data scientist has examined the figures and found that Super Rugby is superior to its European counterparts.
www.smh.com.au
1 2