F. Javier Rubio

Approaches for modelling survival time in groups with very low risk I'm currently working on a study with a group of hematologists. Patients with PV (Polycythemia vera) have very low risk of future thromboembolism after diagnosis due to disease management. Patients

I've got 1 event in one group in which we know the risk of the outcome is very low.

This is creating enormous confidence intervals. Can I use firth penalization? Should I get more data? All ideas welcome

stats.stackexchange.com/questions/66...

4-panel comic. (1) [Person 1 with ponytail flanked by person with short hair and another person speaking into microphone at podium] PERSON 1: In the early 2010s, researchers found that many major scientific results couldn’t be reproduced. (2) PERSON 1: Over a decade into the replication crisis, we wanted to see if today’s studies have become more robust. (3) PERSON 1: Unfortunately, our replication analysis has found exactly the same problems that those 2010s researchers did. (4) [newspaper with image of speakers from previous panels] Headline: Replication Crisis Solved

Replication Crisis

xkcd.com/3117/

Conformalized Regression for Continuous Bounded Outcomes Regression problems with bounded continuous outcomes frequently arise in real-world statistical and machine learning applications, such as the analysis of rates and proportions. A central challenge in...

New preprint with Fabrizio Leisen and our PhD student Zhanli Wu:

"Conformalized Regression for Bounded Outcomes"

arxiv.org/abs/2507.14023

#rstats #conformal #prediction

R code and data are also available at:

github.com/ZWU-001/CPBo...

Research Studentships

🚨3 Departmental PhD Studentships have now become available. The deadline for applications will be 31 of July. Available for overseas and home students.

www.ucl.ac.uk/statistics/p...

The latest issue of the ISBA bulletin, containing a call for contributed discussions for two papers:

5-panel comic. (1) [teacher with long hair next to whiteboard] TEACHER: I’m supposed to give you the tools to do good science. (2) [teacher addressing students] But what *are* those tools? Methodology is hard and there are so many ways to get incorrect results. What is the magic ingredient that makes for good science? (3) TEACHER: To figure it out, I ran a regression with all the factors people say are important: [embedded list in sub-panel, cut off at end] Outcome variable: correct scientific results. Predictors: collaboration; skepticism of others’ claims; questioning your own beliefs; trying to falsify hypotheses; checking citations; statistical rigor; blinded analysis; financial disclosure; open data (4) TEACHER: The regression says two ingredients are the most crucial: 1) genuine curiosity about the answer to a question, and 2) ammonium hydroxide. (5) STUDENT: Wait, why did *ammonia* score so high? How did it even get on the list? LONG HAIR: ...And now you’re doing good science!

Good Science

xkcd.com/3101/

Call for discussion papers 2025: Innovative usages of natural experiments and causal inference in statistics and data science👇

rss.org.uk/news-publica...

a novel discrepancy measure My friend EJ Wagenmakers, along with his colleague  Raoul Grasman, have proposed a novel measure of discrepancy between two distributions that they formulate through a basic Bayesian lens as an exp…

A novel discrepancy measure:

xianblog.wordpress.com/2025/06/01/a...

GitHub - FJRubio67/HazReg: Parametric hazard-based regression models (R package) Parametric hazard-based regression models (R package) - FJRubio67/HazReg

Not a proper ML guy, but I have used this trick a few times:

1. In the log-likelihood of relative survival models.

github.com/FJRubio67/Ha...

2. Some “intractable likelihood” models:

rpubs.com/FJRubio/AMLE...

A nice overview of GAMs:

"Generalized Additive Models" by Simon N. Wood

doi.org/10.1146/annu...

Interesting to see cancer epidemiology papers in the top 10:

3 year PDRA position in the OCEAN project (https://oceanerc.com/) to work at Warwick with Gareth Roberts, Adam Johansen and other OCEAN Researchers on a range of topics around scalable distributed computation. Deadline 10th April 2025 at 11.55pm UK time. Details at
warwick-careers.tal....

Or, to clarify notation:

Archive Request xkcd.com/3052

🎉 Happy 199th birthday, UCL! 🎉

We’re just one year away from celebrating 200 years of innovation, discovery, and impact.

UCL has always been a place of firsts. Now, we look ahead to an exciting bicentenary year!

Learn more: www.ucl.ac.uk/news/2025/fe...

#UCL200 #UCLFoundationDay #UCLBirthday

Hi #stats and #datascience community, we are compiling a list of resources for teaching multivariable thinking. Please feel free to share any resources that you might have developed and/or use in your own teaching. I'd appreciate it if you can help me spread the word.

I'm really looking forward to the CRiSM 2.0 Conference warwick.ac.uk/fac/sci/stat... from 21st-23rd May. We've been lucky enough to get a really nice list of speakers; registrations open now if anyone is in the market for some interesting talks in May.

Principal Investigator positions at ELLIS Institute Finland | ELLIS Institute Finland Now recruiting new PIs in artificial intelligence and machine learning

I have big news: @ellis.eu has launched its 2nd major research center, @ellisfinland.bsky.social! I have agreed to start as founding director & the first call for PI positions is open. This is a major opportunity for outstanding researchers, join us! ellisinstitute.fi/PI-recruit

Problems caused by grade inflation
statmodeling.stat.columbia.edu/2025/01/19/p...

Interesting short paper about the debates on Bayesian inference in the 1950s

Bayesian issues in the 1950s: an episode involving Karl Popper and Jimmie Savage

- Stephen M Stigler

doi.org/10.1093/jrss...

Looks to be a really nice read; clear that the man enjoys writing!

arxiv.org/abs/2501.00925
'Fisher Information in Kinetic Theory'
- Cédric Villani

Diagram with large number: 2.7.123 First “2” is commented: Proud version. Bump when you are proud of the release Second “7” is commented: Default version. Just normal/okay releases Third “123” is commented: Shame version. Bump when fixing things too embarrassing to admit

I propose we replace semantic versioning with pride versioning

Winter 2025 - Researcher positions in AI and machine learning — FCAI

Postdoc and doctoral student positions in developing Bayesian methods! The positions are funded by Finnish Center for Artificial Intelligence FCAI and there are many other topics, too, but if you specify me as the preferred supervisor then it's going to be Bayesian. fcai.fi/winter-2025-...

One #postdoc position is still available at the National University of Singapore (NUS) to work on sampling, high-dimensional data-assimilation, and diffusion/flow models. Applications are open until the end of January. Details:

alexxthiery.github.io/jobs/2024_di...

One of the risks in testing before submitting is that you may end up training chatGPT. So, by the time the assessment is released, the AI may have already learned those questions. It is not clear or certain, though, as we do not know how those AIs are trained.

That is quite good. I know several colleagues who have tested their assessments in ChatGPT and it gets very close to full marks.

Also as Last Author, so it also solves the problem in medical and epi journals!

rspatialdata: a collection of data sources and tutorials on downloading and visualizing spatial data with #rstats 💻🗺️📊

👉 rspatialdata.github.io

Ig Nobel Face-to-Face (video part 2 of 3) The Ig Nobel Prizes honor things so surprising that they make people LAUGH, then THINK. At the 2024 Ig Nobel Prize ceremony, ten new prizes were awarded. Two days later, most of&nbsp…

Video of the Ig Nobel Face-to-Face meeting, which allowed time for discussion. I'm talking about our 350,757 coin flips. Also features the UvA work on drunken worms.
improbable.com/2024/12/03/i...

I can't see why not. Priors and Posteriors quantify uncertainty, it does not mean the true generating model is changing every time.