Just boarded a train with my bike in Berlin. If all goes smoothly I should reach Montpellier within 21 hours, with plenty of time before the launch of #ecobhvr2025. Only one change in Paris... So, time to unpack the laptop and check my slides (perhaps after an episode of Wednesday #2...).

Do Spix's macaws really behave as they do in the animated films Rio? To find out, read our new preprint: doi.org/10.1101/2025...

Landed in Budapest. About to spend a week with the Finnish national team of horseback archery! 🏹🐴🇫🇮

I've just created a feed called "wildlife photography".
It catches posts with pictures that are about wildlife through a collection of hashtags that I will continuously refine to improve the feed.
It was easy thanks to skyfeed.app
Please pin/save and let me know what you think.
#wildlifephotography

A more simple illustration of the same #rstats coercion oddity 👇

Visiting the website of Even Rouault (a core developer behind GDAL and other spatial libraries used by many packages in #rstats, #python...), I bumped into this mini opinion about "Fighting misconceptions about Open source": www.spatialys.com/en/opinions/

I just rediscovered that `lintr::lint_package()` is a great solution to check #rstats code in your package.
lintr.r-lib.org

Just discovered the root of a #rstats bug emerging only every 100,000 iterations 👇... (coercing is always dangerous, but I had not seen that one coming)

My fresh trials confirms this (see 👇, the red is the expectation).
Depending on sample size, parameters, and runs, it can be quite far from perfection however.
This is I believe a limitation of smoothers in general.
I am planning to do this: github.com/courtiol/tim... to help.

Tempted to measure how the correlation between 2 variables varies over time?

If the answer is YES, you may find my #rstats 📦 **timevarcorr** (newly on CRAN) useful: courtiol.github.io/timevarcorr/

I implemented the method described in this paper: link.springer.com/article/10.1...

Enjoy!

If you develop #rstats 📦 and don't know about r-universe.dev , give it a try, it is really good!
For example, r-universe.dev/search/ allows you to identify the #rstats 📦 you contributed to.
They have even created a page for every maintainer listed by CRAN, e.g. for me: courtiol.r-universe.dev

Have you ever found #rstats rowwise computations cumbersome, slow, or memory greedy?

If so, do give a try to **lay** -- a small 📦 from @romainfrancois.bsky.social and myself.

Check courtiol.github.io/lay/ for examples, alternatives and benchmarks.

You can use **lay** alone, or with **dplyr** 👀👇

Embracing FOMO: my first trial of GitHub AI Co-pilot integration within @posit.co RStudio for #rstats...

Will I hate it or love it? 🤔

Check out the update on the planetary boundaries framework, which evaluates 9 processes "critical for maintaining the stability and resilience of Earth system":
www.science.org/doi/10.1126/...

We had entered the danger zone for 3 of the 9 processes by 2009, 4 by 2015, and the new number is 6 🫣

I just fixed a corrupted docx file by hand 🥳. I unzipped the docx, opened word/document..xml, loaded the file in jsonformatter.org/xml-viewer, beautified it so as to unpack the megaline of XML code into chunks, removed the troublesome blocks, zipped everything back and renamed as docx, et voila!

You live in #Berlin? You like Tempelhofer Feld? --> sign the petition to make it a UNESCO world heritage site: weact.campact.de/petitions/te...

I just managed to use the domain name I had already purchased from GoDaddy as my handle. In GoDaddy, I selected DNS records > Add new record > Type = TXT, Name = _atproto, did=did:XXXXX (whatever Bluesky tells you in Settings > Advanced > Change handle > "I have my own domain". Easy!!