I used to learn everything i know about statistics on this website, these days I only learn what’s between any combination of letters on my keyboard

NEW PREPRINT

The harm of the use of imbalance correction methods on calibration, extending our earlier work on regression models to more complex techniques
arxiv.org/abs/2404.19494

”Solving” imbalanced classes “problem” has been a flourishing cottage industry academic field.

Perusing papers in imbalanced classification has been a real eye opener, one would have thought SMOTE paper with 25K citations is an exception rather than a rule.

Nah. There is a…

I've long advocated against the use of SMOTE for correcting class imbalance, observing that it can disrupt model calibration.

My book, 'Practical Guide to Applied Conformal Prediction in Python: Learn and apply the best uncertainty frameworks to your industry applications',…

Was reminded again of these simple rules to academic success today

lol who did this

🚨 NEW OPPORTUNITY: SPECIAL ISSUE 🚨

Special issue on the validation and transparency for AI-based prediction models. Welcoming research articles, methodology papers, reviews, protocols, and commentaries.

Feel free to contact me for more info
biomedcentral.com/collections/VT…

A methodologist trying to fix measurement issues in the analysis stage.

Kind reminder: your algorithm is dumb as f*ck

David Hand's 'Classifier Technology and the Illusion of Progress' from 2006 is still one of the most useful papers I use in peer review. IMHO, it is a must read for anyone comparing algorithms (stats vs ML, anyone?) on their datasets

Open access: projecteuclid.org/euclid.ss/1149…

In case you missed it, already forgot about it or if you are desperately looking for reading material for the weekend: TRIPOD+AI came out this week 👇

bmj.com/content/385/bm…

👇👇👇

This, and also for journal styles, promotion documents, reference letters (or better: just stop with that), etcetera

NEW PAPER in Journal of Clinical Epidemiology 'SPIN-PM: A consensus framework to evaluate the presence of spin in studies on prediction models'

--> tinyurl.com/4nwa4eb6

#machinelearning #spothehype #seethroughthehype #keepitreal

NEW PAPER out today in @BMJ_latest TRIPOD+AI: reporting recommendations for studies developing or validating prediction models for use in healthcare that use #machinelearning methods

bmj.com/content/385/bm…

#ArtificialIntelligence #AIstandards #OpenAccess

Please share 🙏

Long anticipated, it's here now