Andy Seaton

New paper: inlabru: software for fitting latent Gaussian models with non-linear predictors Preprint now up on arxiv arxiv.org/abs/2407.00791 Inlabru extends the class of models that users can fit with INLA (1/n)

@dan_p_simpson @ArtisanalSudoku Renban in box 8* had the 6 pencil marks I meant

@dan_p_simpson @ArtisanalSudoku Ah I did wonder about the name. I was looking at renban in box 7 with either 1 on it or off it, felt that 1 on it looked like it would be the way so wanted to break 1 off the renban. It does break using the 6 on the renban in box 7 but I couldnt see it without pencil marks.

@dan_p_simpson @ArtisanalSudoku Nice! I liked the break in. The nasty bit did keep me for long enough that I'm not sure I solved it in the intended way, I felt like I was on the verge of bifurcating.

@BenCAugustine In that case I agree with your intuition. I find it quite hard to think about how things can go wrong when it could be some combo of how traps are placed in relation to the habitat (preferential sampling problem) and/or a poor detection model (mis-specification problem)

@ASeatonSpatial Actually, we did smooth the habitat covariate at the scale of female home ranges here.

Recently reminded of this case where SCR activity centers in fragmented landscape get "pushed" into poor habitat and flip the expected sign of the density beta. My top hypothesis was no traps in poor habitat + unmodeled detection heterogeneity. Just encountered this again.

@BenCAugustine Reasoning something like: the activity centre is just representative of a centre of mass and the important thing is the covariate values in a region around that point.

@BenCAugustine I have often wondered whether we should give some advice like "kernel smooth your covariates" and possibly for that smoothing to take into account movement model (or detection model however you want to view it). So using e.g. half-normal sigma to inform bandwidth of the smooth.

This made me remember a British Ecological Society event for early career researchers I was at once where the keynote speaker gave a talk that was along the lines of "this profession can sometimes feel like doing an autopsy on a dying world, but don't lose hope!"

From: academic.oup.com/jrsssb/article… Reading this prompted by reading a paper where DIC selection does a lot of heavy lifting re "real" or "not real" effects.

Premium sass levels reached

@TanqsAndTonic @FraserNelson If you click through to the dashboard I think that it refers to the difference with other charts in the dashboard that have figures like 227 per million instead of 277. Depending on the source of the data for each chart not all of them report the exact same incidence.

The Spectator data hub now has a special section devoted to Scotland's (staggering) drugs crisis. data.spectator.co.uk/drugs-crisis

@GivingTools Okay this looks cool, I had never heard of this.

@ASeatonSpatial Solution: no 'publications in journals'. Just commission to have work (e.g., a Quarto/notebook) publicly evaluated, rated, & reproducibility checked at a point in time. Authors can then improve it in situ & seek another rating round. #unjournal.org

I can't wait until we move away from the idea that a paper is a static pdf that has gone through a review and instead treating it more like a completely open software product that people can submit bug reports to and can evolve over time without hard to find eratta

@MomentumOp89 @zhil_arf Yeah we do that, and its basically already the status quo (number of papers published per day far exceeds number read). People will still find good stuff, bad stuff will be more openly bad. I just see no point in relying on 1-3 people doing secret stuff we never get to see.

@ASeatonSpatial @zhil_arf but if scientific research is “open source” before it is let loose in the public and opened source it should at least be verified for a certain level of credibility. Otherwise do we just spam our latest research interests onto the internet? (something I’m not against btw)

In other fields it probably makes less sense to think this way.

Obligatory caveat: this is what I think about my field (statistics) where often large amounts of bespoke code is involved in producing the key findings of the paper

@AGoldmund For consistency we'll need a version of We Three Lightsabers

@undertheraedar I appreciate the attention to detail since of course the transportation happened before that bridge became bus + taxi only.

the last sentence from the paper is really mind-blowing, or maybe I just don't think about Neolithic transport networks enough "A Scottish provenance for the Altar Stone implies Neolithic transport spanning the length of Great Britain."

Today's Stonehenge news is weirdly exciting and it's the most excited I've been about a 6 tonne shaped block in quite some time, plus you can get the actual paper pdf by simply clicking a link - so gneiss to sea! nature.com/articles/s4158…

📖Published📖 A Monte Carlo resampling framework for implementing goodness-of-fit tests in spatial capture-recapture models 👇 doi.org/10.1111/2041-2…