| rud.is | https://rud.is/ |
| observable | https://observablehq.com/@hrbrmstr |
| github | https://github.com/hrbrmstr |
arXiv is freshly hiring for 3 positions:
- Software Engineer
- DevOps Software Engineer
- Software Engineer Scientist
US-only, NYC-based, hybrid/remote possible.
Share and help us build a backbone of Open Science.
If you want to start or get better at data visualization in 2024, you should def take advantage @andykirk's few-remaining open slots in his inaugural 2024 'Fundamentals' course.
Zero pre-req’s.
https://visualisingdata.com/2023/09/fundamentals-of-data-visualisation-jan-2024/
So, I don't 👀 at "follower" (terrible term) lists b/c narcissism corrupts & why repeat the mistakes of Twitter/X?
That means I don't know who from fosstodon follows this acct.
So, I'm letting said fosstodon-ers know I asked them to ban this acct since they banned my primary @hrbrmstr acct in Oct.
I wld not want to accidentally offend their sensitive sensibilities via this one, too.
Will be banning the fosstodon domain from this acct as well after a short delay.
Was nice chatting w/y'all!
What a roller coaster of a week between manageable and ugh days.
Delighted I've got the new R WASM toy to keep me focused on something when not engaged in work or fam stuff, tho. Esp when this thing decides to not let me sleep.
I haven't spammed that R stuff here (I don't think I have). It's more "tech" than vis, but recent stuff has focused on vis.
A look at tagged, malicious traffic. Link to the blog post featuring these Tagged Malicious Traffic Started Coming In As Soon As The Sensors Were Functional 217,852 total malicious events encoutered during the ~7.8 day sampling period. The Four Largest "Spike" Hours Had Mostly Similar Characteristics August 28 malicious traffic focused mainly on SMB exploits, and originated from the Data Communication Business Group autonomous system in Taiwan We Saw The Usual Suspects Rise To The Top Of 13,576 Ports Telnet
In other good news, I used @observablehq for all the vis in a blog post coming out later today (will link when it's out). The provider we uses makes embedding the non-Observable branded iFrame's oddly hard (stuff comes out weird). If Observable had a white-on-black version of the bottom row branding, I likely could have used it.
I also need to provide feedback abt SVG and PNG exports not working well, but that's not happening today.
Sunday was "good", relative to actually contracting this bugger.
Monday was…not.
Today, back to the Sunday "good" level.
Taking the small wins with serious appreciation. I cannot imagine what it was/is like for the folks who didn't/don't have "good" days during this.
📣 I've got a new #preprint out, with @juliasilge , @hfrick , @topepo , plus Lucas Johnson and Colin Beier!
We give a head-to-head comparison of spatial cross-validation methods, give advice on applying spatial CV for applied modeling projects, and walk through how these techniques are implemented in the {spatialsample} #rstats package.
Preprint: https://arxiv.org/abs/2303.07334
Repo: https://github.com/cafri-labs/assessing-spatial-cv
Evaluating models fit to data with internal spatial structure requires specific cross-validation (CV) approaches, because randomly selecting assessment data may produce assessment sets that are not truly independent of data used to train the model. Many spatial CV methodologies have been proposed to address this by forcing models to extrapolate spatially when predicting the assessment set. However, to date there exists little guidance on which methods yield the most accurate estimates of model performance. We conducted simulations to compare model performance estimates produced by five common CV methods fit to spatially structured data. We found spatial CV approaches generally improved upon resubstitution and V-fold CV estimates, particularly when approaches which combined assessment sets of spatially conjunct observations with spatial exclusion buffers. To facilitate use of these techniques, we introduce the `spatialsample` package which provides tooling for performing spatial CV as part of the broader tidymodels modeling framework.
I 💯% had 💯% empathy for anyone who contracted covid since the start of the pandemic, regardless of “why”.
Now that I am clearly in the throes of a “who knows how” long covid experience — despite taking inanely draconian precautions — please know *you* are an incredible human, & that anything you are able to do, each day, mattersl & matters dearly. Even if said “thing” is “waking up each day”.
Hang in there.
We (which inherently means *you*) — collectively – “got this”.
Deyan Ginev
Mike Mahoney