I fell in love with the R ecosystem during my PhD. Thousands of open source packages, all free. I never had to depend on Microsoft or a university license to do science. Now I write ecology packages and put them on CRAN. Independent tools for independent science. #rstats
Remote
Gilles Colling
@GillesColling@fosstodon.org
mastodon 4.7.320 Followers
19 Following
17 Posts
Joined February 25, 2026
left_join() with relationship = "many-to-many" suppresses the warning. But a warning you can silence isn't a contract.
join_strict() makes cardinality explicit: 1:1, 1:many, many:many. State what you expect. Get a hard error when it's violated.
Replying to
Fix: data.table::set() modifies in place, or vectorize so the modification lands in one shot. #rstats 2/2
New alien plant species per decade in Austria:
1950s → 14
2010s → 47
No plateau. The rate itself is accelerating.
VIF says drop temperature. But there are 23 valid subsets of uncorrelated variables — and temperature survives in 19 of them.
VIF picked one subset and hid the other 22.
corrselect enumerates ALL of them. You choose.
80% of eBird checklists in Austria are within 500m of a parking lot.
That species gradient you're mapping? It's a gradient of where people stop their cars.
Relative abundance went up 30%.
Sounds like good news — until you realize the focal species stayed flat and everything else crashed.
A ratio can't tell you which side moved. numdenom models numerator and denominator jointly so you actually know.
Replying to
Transitive dependency effects are underappreciated. One package updating a shared dep can ripple through dozens of reverse dependencies without anyone noticing until CRAN check day. Good write-up on making the downstream effects visible.
Replying to
But if your segfault frequency is climbing — meaning your data structures have complex ownership where memory gets accessed from multiple places — rextendr eliminates that entire class of bugs before your code ever runs. #rstats 2/2
Replying to
If accumulation deviates from what random placement would produce, there's spatial structure the decay curve missed — and you can see exactly where it breaks down. #ecology 2/2
Relative abundance went up 30%. Which part moved — focal species increased, or everything else decreased?
A ratio collapses two biological processes into one number. Can't tell which drove it.
numdenom fits numerator and denominator as a joint model with shared random effects. Sites that are good for the focal species tend to be good overall — the model captures that correlation instead of collapsing it. The ratio becomes interpretable again.
Square grid cells at the equator are 2x the area of cells at 60°N. Bigger cells capture more observations — and if a spatial process spans more than a few degrees of latitude, that area distortion might mess up ALL the downstream statistics.
hexify uses the Snyder equal-area projection — a polyhedral projection that tiles the Earth into hexagons where every cell has the SAME area regardless of latitude. C++ backend, scales from 10 km² field plots to 7,774 km² continental grids.
Replying to
Rust is the right call for a linter. Fast enough to run on save without anyone noticing. How does Jarl handle tidyeval? That's where R linters usually break down, either flagging every rlang pattern or ignoring them completely.
Replying to
Rapid green infrastructure assessment for Calgary communities — that's a practical tool. If you're working with grid-based spatial data, equal-area hexagonal grids prevent the area distortion that lat/lon cells introduce at higher latitudes. hexify does this with a fast C++ backend.
https://cran.r-project.org/package=hexify