@zeileis@fosstodon.org
@zeileis@fosstodon.org avatar

zeileis

@zeileis@fosstodon.org

Statistics, science, software.

This profile is from a federated server and may be incomplete. Browse more on the original instance.

zeileis, to random
@zeileis@fosstodon.org avatar

Very sad news: Our friend and colleague Fritz Leisch passed away this week.

Among his many contributions:

  • Original R Core Team member
  • Co-founder of CRAN
  • Sweave author for reproducible documents

R Core Team announcement:

https://stat.ethz.ch/pipermail/r-announce/2024/000702.html

BOKU obituary (in German):

https://boku.ac.at/news/newsitem/78553

leonido, to random German
@leonido@chaos.social avatar

This week I had the great honor of providing the opening keynote to the conference "Enhancing the voice of science on Wikipedia: How universities can collaborate with the online encyclopedia in science communication" hosted at @uniinnsbruck

The video recording of my talk and the slides are already online available, please check them out over at @osconjunction

https://osconjunction.net/2024/04/11/opening-keynote-on-science-communication-and-wikipedia-potentials-and-pitfalls/

zeileis,
@zeileis@fosstodon.org avatar

@leonido @uniinnsbruck @osconjunction Great opening lecture, I very much enjoyed it!

Regarding YouTube & video licenses & reuse in Wikipedia: Is it good enough to flag your YouTube videos as CC-BY on YouTube (instead of the default YouTube license)?

That's what I do with my own YouTube videos (https://www.youtube.com/@azeileis) but I wondered whether I should do something else instead of or in addition to that?

gmcd, to random
@gmcd@mastodon.social avatar

Stoked to announce a new release of plot2, the powerful and lightweight (0-dep) extension 📦 for base plotting!

What's new? A bunch of things, including area plots and facets. Draw plots like the below with single/simple function calls. https://github.com/grantmcdermott/plot2

zeileis,
@zeileis@fosstodon.org avatar

@devSJR @gmcd Thanks for the pointer, I have updated the reference in the tutorial now.

https://github.com/grantmcdermott/plot2/pull/114

brodriguesco, to random
@brodriguesco@fosstodon.org avatar

My biggest regret of my short academic career is this paper that I got published:

https://link.springer.com/article/10.1007/s10614-015-9513-8
Not a great paper, But at the time and was not quite known among economists

But that’s not the problem, the problem is that it’s behind a paywall!

zeileis,
@zeileis@fosstodon.org avatar

@brodriguesco @RPanczak I'm fairly certain that the copyright transfer agreement that Springer used at the time already allowed self-archiving.

What I do in those cases is take the last sources, use my own style files, and put a footer with copyright notice, DOI, and reference into it. For a 2015 example, see:

https://www.zeileis.org/papers/Strobl+Kopf+Zeileis-2015.pdf

Some journals might have had a self-archiving embargo for a limited amount of time but (a) I routinely ignored this and (b) by now the embargo will have passed.

zeileis, to random
@zeileis@fosstodon.org avatar

PSA (in German)

Mathematik für Wirtschaftswissenschaften: Eine Einführung mit anwendungsorientierten Beispielen

https://www.mathe4wiwi.org/

Neue Lernplattform mit Onlinebuch (+ Print), Videos, Folien, interaktive Trainingsaufgaben.

1/

andrew, to random
@andrew@fediscience.org avatar

Because of Reasons I have to calculate predicted values from a logistic regression model using manual matrix multiplication isntead of predict() and I can never remember how to do it correctly, so I wrote myself a tiny little guide for it. https://www.andrewheiss.com/blog/2023/08/15/matrix-multiply-logit-predict/

zeileis,
@zeileis@fosstodon.org avatar

@andrew I would recommend to build the model matrix with the formula/terms stored in the model itself. This makes sure it is in sync with the model coefficients.

x <- model |>
terms() |>
delete.response() |>
model.matrix(data = data_to_plug_in)

Then you can compute the linear predictor as before:

lin_pred <- drop(x %*% coef(model))

where drop() preserves the row names.

Finally, the inverse link is also stored in the model object:

model$family$linkinv(lin_pred)

zeileis,
@zeileis@fosstodon.org avatar

@andrew brms has its own formula/terms because it needs more types of terms. So it is possible to do something like:

brmsterms(model$formula)

But I'm not sure what is the recommended way to go about handling these in packages that build on top of brms.

ijlyttle, to random
@ijlyttle@vis.social avatar

Very excited to participate on Haley Jeppson’s panel for “Use of Color in Statistical Charts” at ! (Sunday 2-4 PM, CC-810)

I’ll use this (in progress) collection of @observablehq notebooks for computing on color:

https://observablehq.com/collection/@ijlyttle/color

Thanks to @zeileis, whose colorspace package in R inspired much of this work:

https://hclwizard.org/r-colorspace

zeileis,
@zeileis@fosstodon.org avatar

@ijlyttle @observablehq Very nice, impressive collection of useful tools! The interactive widgets are particularly fun to play around with.

zeileis,
@zeileis@fosstodon.org avatar

@ijlyttle @observablehq Re: Unresolved thoughts on color vision deficiency.

In addition to the model-based implementation in {colorspace}, there is the classic {dichromat} based on experimental work by Brettel, Mollon, and Viénot in the late 1990s.

You can use these to check the differences.

https://CRAN.R-project.org/package=dichromat

zeileis,
@zeileis@fosstodon.org avatar

@ijlyttle @observablehq Comparison of deutan emulations:

library("colorspace")
library("dichromat")
vi <- hcl.colors(9)
rb <- rainbow(9)

swatchplot(list(
"viridis" = rbind(
original = vi,
colorspace = deutan(vi),
dichromat = dichromat(vi, "deutan")),
"rainbow" = rbind(
original = rb,
colorspace = deutan(rb),
dichromat = dichromat(rb, "deutan"))),
nrow = 4)

zeileis,
@zeileis@fosstodon.org avatar

@ijlyttle @observablehq Yes, I can very much relate to that! This kind of curiosity inspired a lot of the functionality in colorspace.

I'm very happy to see that this was useful to you and that you took everything a couple of steps further.

djnavarro, to random

Today, thanks to a random github comment by @grrrck, I discovered that has pre-render and post-render options that allow you to execute custom scripts before and after rendering a project. Just add something like this to the _quarto.yml file and... it works 🎉

project:  
 type: website  
 post-render: ./post-render.sh

I'm way too excited by this, aren't I? Yeah, even I can see that

zeileis,
@zeileis@fosstodon.org avatar

@grrrck @djnavarro Just what I needed for my book project, thanks for sharing this!

Is it possible to specify in the .yml to only post-render the "html" target?

At the moment I need to catch the "pdf" and "epub" targets in the shell script because only the "html" needs improvements.

zeileis,
@zeileis@fosstodon.org avatar

@MickaelCanouil That's really handy, thanks for the link. Based on this I could also simplify the script itself.

@grrrck @djnavarro

brodriguesco, to random
@brodriguesco@fosstodon.org avatar

New blog post: How to self-publish a technical book on Leanpub and Amazon using Quarto

https://www.brodrigues.co/blog/2023-06-29-book_quarto/

zeileis,
@zeileis@fosstodon.org avatar

@brodriguesco This is perfect and comes exactly at the right time for me. Thanks, Bruno!

We are just wrapping up a quarto book and I was wondering how to do these things.

  • Web is almost done.
  • PDF is ok but I will profit a lot from your advice.
  • Epub didn't work yet because of some math array stuff that I wasn't able to figure out yet.

I have a couple of follow-up questions, though, if I may. They are hopefully short/easy...

zeileis,
@zeileis@fosstodon.org avatar

@brodriguesco

  1. If I want to sell the printed book and the epub, I would have to set up separate deals with Amazon and Leanpub, respectively?

  2. If I just want to sell the printed book, then only a PDF for Amazon is needed?

  3. How does the cover work for the printed book, is there some template for it?

  4. What is the difference/advantage of documentclass "scrbook" vs. "scrreprt"?

zeileis,
@zeileis@fosstodon.org avatar

@brodriguesco Very much appreciated, thanks for the advice!

Possibly we will then start with Amazon only bc I don't know whether I will get the epub ready before the teaching (based on the book) starts.

How long is the delay between uploading the PDF and Amazon shipping copies?

zeileis,
@zeileis@fosstodon.org avatar

@brodriguesco But that's still very fast. If I introduce it in one week, they can all have the book for the second lecture 😇

zeileis,
@zeileis@fosstodon.org avatar

@brodriguesco Thanks for the pointer. That's very useful in general, I'm not sure whether it will be the solution here. It's a 500-page introductory math book, originally written fully in LaTeX.

Getting everything into Quarto with HTML and PDF in sync was already a bit of work...

zeileis, to random
@zeileis@fosstodon.org avatar

New working paper with Paul Murrell.

Overview and advice for new color palette functions palette.colors() and hcl.colors() in base since version 4.0.0.

Short blog post:

https://www.zeileis.org/news/coloring/

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • normalnudes
  • hgfsjryuu7
  • magazineikmin
  • thenastyranch
  • Youngstown
  • slotface
  • everett
  • ngwrru68w68
  • mdbf
  • kavyap
  • tsrsr
  • Durango
  • PowerRangers
  • DreamBathrooms
  • Leos
  • InstantRegret
  • khanakhh
  • osvaldo12
  • vwfavf
  • tacticalgear
  • rosin
  • cubers
  • cisconetworking
  • GTA5RPClips
  • ethstaker
  • tester
  • modclub
  • anitta
  • All magazines
    •