Matt Kmiecik

My favorite figures (to date) In my previous post, I discussed my pipeline for making publication-ready figures.

Maximizing information while minimizing clutter is key to a great figure. In this post, I cover my approach for various figures throughout the years.

Read more: mattkmiecik.substack.com/p/my-favorit...

#rstats #dataviz #science #publishing

Certainly! I wouldn't recommend going to GUI if the data will change.

Yes! Don't get me wrong, everything is possible in ggplot2 and it's extensions. But I've found it nice to polish off figures in a GUI when at the final stages or experimenting with different aesthetics/placement, etc.

Using Omnigraffle for publication ready figures In graduate school, I took an advanced statistics course that changed my life.

ggplot2 is powerful, but not always pretty. 😬

That’s why I use Omnigraffle for final figure touches—cleaner legends, custom labels, and full control over design.

Full workflow here: mattkmiecik.substack.com/p/using-omni...

#rstats #dataviz #omnigraffle #ggplot2

Scaling variables in R A walkthrough of a tip I recently learned

Just dropped a new #rstats Substack post!

It’s about a better way to scale variables in R that won’t break predict() (and keeps your model inputs traceable).

mattkmiecik.substack.com/p/scaling-va...

#Tidyverse #DataAnalysis

Great, thanks!

Is there a formal way to give yourself permission or is this implied by CC-BY?

Scaling variables in R A walkthrough of a tip I recently learned

Z-scoring in R? There’s a neat way to retain the original mean and SD after using scale(). This makes back-transforming for plots super easy.

Full example: mattkmiecik.substack.com/p/scaling-va...

#rstats #Reproducibility #CodingTips

I started this Substack just so that I could use Ulysses Alright, that might be an exaggeration.

Scientific writing has trapped me in Word & Google Docs. For personal writing, I needed something better. Ulysses is that something. Clean, simple, and dare I say... fun?

Here’s why I love it:
✍️ mattkmiecik.substack.com/p/i-started-...

#WritingCommunity #TechTools #MacApps #ProductivityHacks

🚨 Big milestone for Rdatasets 🚨

The web archive now hosts 3400+ free and documented CSV datasets. Fantastic for teaching and testing!

And {Rdatasets} is a new #RStats 📦 for easy download and search

Web archive: vincentarelbundock.github.io/Rdatasets

R 📦: vincentarelbundock.github.io/Rdatasetspkg

Scaling variables in R A walkthrough of a tip I recently learned

I used to wrap scale() in as.numeric() and call it a day. Now I know better: drop() preserves attributes and behaves cleanly.

Full walkthrough (with mtcars, lm, ggplot2): mattkmiecik.substack.com/p/scaling-va...

#rstats #DataViz #OpenScience

Personally I prefer using datawizard::standardize(), and specifically using it *in the formula*.

So
mtcars$hp_z <- scale(mtcars$hp)
mpg ~ hp_z

Becomes
mpg ~ standardize(hp)

This solves both issues you raise in your post.

#rstats @easystats.bsky.social

Nice! Thanks for commenting about datawizard::standardize(). It looks like a great solution 😀

Scaling variables in R A walkthrough of a tip I recently learned

Scaling variables in R? scale() is powerful — but its default behavior can mess with predict().

Here’s a clean fix using drop() + how to retain scaling attributes for later.

mattkmiecik.substack.com/p/scaling-va...

#rstats #DataScience #Statistics

GitHub - tidyverse/ragnar Contribute to tidyverse/ragnar development by creating an account on GitHub.

github.com/tidyverse/ra...

This has to be the best #rstats hex sticker I've ever seen

My start with R The programming language that made me a better scientist

Here's one in the other direction! mattkmiecik.substack.com/p/my-start-w...

nice

Too funny! 🤣

Labeling facets in ggplot2 Who knew such a small setting could make such a big improvement

One of my favorite tricks I’ve learned from ChatGPT: better facet labels in ggplot2.

Check out the latest post: mattkmiecik.substack.com/p/labeling-f...

#rstats #ggplot2 #dataviz

Labeling facets in ggplot2 Who knew such a small setting could make such a big improvement

Adding labeller = label_both to your facet calls in ggplot2 can massively improve figure readability! I wrote a quick blogpost with a use case and some examples.

mattkmiecik.substack.com/p/labeling-f...

#rstats #ggplot2 #VisualizationTips #dataviz

Most applications in R run pretty quickly. But these days, many R users are faced with huge datasets and/or long-running applications, making parallel computation attractive. 🧵 1/

Labeling facets in ggplot2 Who knew such a small setting could make such a big improvement

Struggling with confusing facet labels in ggplot2? I just shared a quick fix that makes your plots clearer using labeller = label_both.

mattkmiecik.substack.com/p/labeling-f...

#rstats #dataviz #ggplot2 #DataScience

Named vectors in R A powerful tool for plotting colors

R is full of surprises—even after a decade, I’m still learning simple but powerful tricks. Named vectors are my latest discovery, thanks to ChatGPT!

mattkmiecik.substack.com/p/named-vectors-in-r

#RStats #Learning #ChatGPT #Dataviz

Thank you so much Ben! I'm looking forward to giving these a read

Will do! Thanks again

Thanks Norm! I recently discovered the Tweedie distribution; this has really improved things for some of the data I am working with

Great, thank you! Would you recommend any texts about what sort of distributions are available and how to interpret the resultant coefficients?

I'm looking for literature recommendations on guidance for specifying distributional assumptions in generalized linear/additive models. I've mainly stuck with gaussian and binomial distributions, but am looking to explore other distributions to enhance specification. Thank you! #rstats