Skip to content
PodSubs
ShowsPeopleGuestsJobsNewsEvents
More
BlogPodCred
Menu
ShowsPeopleGuestsJobsNewsEventsBlogPodCredNetworksSign in
Join nowSign in
  • Shows
  • People
  • Jobs
  • News
  • More
    • Guests
    • Events
    • Blog
    • PodCred
    • Networks
    Join nowSign in

R Weekly Highlights

R Weekly Highlights artwork

Issue 2026-W45 Highlights

Nov 6, 202644 minAudio
ShareXLinkedInFacebookWhatsAppEmail

What happens when that online data set isn't so easy to download anymore? We learn of two packages to ease the pain of obtaining these files and then some. Plus a visit to the visualization corner with a handy way to send interactive charts without requiring a server-side process, plus the key questions and principles to create amazing plots. Episode Links This week's curator: Jonathan Kitt - @jonathankitt.bsky.social (Bluesky) Introducing safeconduct and r2parquet How to make interactive charts you can email On message and on brand: conversation-starting dataviz with R and ggplot2 Entire issue available at rweekly.org/2026-W45 Supplement Resources Brad Lindblad family fundraiser https://give.lendahandup.org/fundraisers/brad-h-lindblad Supporting the show Use the contact page at https://serve.podhome.fm/custompage/r-weekly-highlights/contact to send us your feedback R-Weekly Highlights on the Podcastindex.org - You can send a boost into the show directly in the Podcast Index. First, top-up with Alby, and then head over to the R-Weekly Highlights podcast entry on the index. A new way to think about value: https://value4value.info Get in touch with us on social media Eric Nantz: @rpodcast@podcastindex.social (Mastodon), @rpodcast.bsky.social (BlueSky) and @theRcast (X/Twitter) Mike Thomas: @mike_thomas@fosstodon.org (Mastodon), @mike-thomas.bsky.social (BlueSky), and @mike_ketchbrook (X/Twitter) Music credits powered by OCRemix The Art of Zoning Out - Pokemon Scarlet - timaeus222 - https://ocremix.org/remix/OCR04570 The Unnamed Frontier - Metroid II: Return of Samus - Pyro Paper Planes, Viking Guitar - https://ocremix.org/remix/OCR02892

Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.

Transcript

From the show. Tap a time to play from there.

Hello, friends. We are back of episode 247 of the RWKY highlights podcast. We've hit November, folks, and my goodness. Yeah. It's hard to believe the end of year is almost here. I think many of us would probably be pretty happy with that. Others might be thinking where the time went, maybe somewhere in between. But where the time is now is you're listening to the weekly ish podcast where we talk about the excellent highlights and, if time permitting, other great resources that are shared in this week's issue of R Weekly,

which you can find at rweekly.org. My name is Eric Nance. I'm delighted that you're joining us from wherever you are around the world. And, yeah, for those that are curious, the Tooth saga is not resolved yet, but, hopefully, will be in the middle of the month. But my rest of my teeth are intact. I had some good candy after the Halloween treat... Trick or treating, but I'm not alone here. Of course, I got my awesome cohost, Mike Thomas, with me. Mike, how are you doing today? I'm doing pretty well. Did we get any listener hate mail last week after our quite unhinged episode or are they still gonna let us podcast this week? Yeah. So far no other cease and desist from our listeners nor the the overlords at Apple and and whatnot. So I think we're still okay for now, but you never know.

Our weekly highlights. That's alright. Crazy. That's alright. You never know what you're gonna get, like a box of chocolates as the famous movie would go, but we got we got a great selection, I dare say so, of highlights today, so we're gonna get right to it and mention, of course, that each issue is a different curator, and this issue was curated by Jonathan Kitt. And as always, had tremendous help from our fellow RWiki team members and contributors like all of you around the world with your poll requests and other wonderful suggestions.

And, Mike, a virtue that we have always advocated for, much in the spirit of R itself, is the access to open data. So many influential things can be done. Whoever you are just learning in data science and you want practice your skills on some real tangible data that's outside that textbook, or if you are trying to improve a process, improve the understanding of a key issue, using open data is a wonderful way to do that and to share those results with the world.

Not always is everything smooth in this aspect and sometimes things go kind of weird when something you could access from say a day ago, a week ago, a month ago, something changes. And so you gotta figure out how to how to account for these changes. And as a data scientist, do you really have time to go under under those rabbit holes, if you will, to solve this? No. But luckily, we have the R community to solve a lot of those issues for us, and that's what this first highlight is gonna do very well here.

And this, blog post introducing not one but two R packages that we're gonna cover here in this highlight was authored by Jens von Bergmann. Hopefully, I got that right. He is a data scientist at Mountain Math Software based in Canada, and shout out to our listeners in Canada, but that's actually a key focus of this blog post because Jen talks about accessing openly available data from the

government of Canada, in particular census data, as well as additional data specific to the city of Vancouver. And in the past, he could get to these via APIs. Well, not anymore because apparently, the organization, putting out these data for Canada called STAT CAN, it started to block more programmatic access to the census data since, earlier this year in October. Now, of course, you can go to the webpage,

click around, download that export manually, but man, when you've had that programmatic access, it's hard to go back to that. And also the city of Vancouver has also changed things a bit where now their API doesn't actually give you the data, it just gives you links to download the data. Figure that one out, folks. I don't know. I'm not gonna pretend to read minds there. So in either case, things are a little more difficult for him to access his data. So he took matters in his own hands and he has two R packages to account for different aspects of this data, importing and data processing.

I'm going to tackle the first one here, Mike will tackle the second one, and the first one is called Safe Conduct, which is basically, if you want to think about all the different ways you can download things on the internet, safe conduct is really meant to abstract away you trying out different methods when your primary method didn't work. I have strong sympathy for this perspective, albeit for a much different use case.

When internally at the day job, we were starting to use R more, but was still dealing in the world of cough cough SAS with SAS seven b data files, and there were times when the Haven package just could not read that file effectively enough, and I had to write an internal shim to do basically two different methods to import that data. Not important what those methods were. They were kind of cludgy at best, but this package feels very similar to that because

this package is gonna go through almost like four or five different methods of downloading a given resource that you pointed to. First, it's going to try a simple curl request. Curl is like a lower level system library to programmatically grab the contents of a link and bring that content down to your system. It's used under the hood practically every software we use that deals with the Internet uses CURL in one way, shape, or form.

If that doesn't work, then it starts to get a little gnarly, albeit in in a very geeky way. It then tries to launch a headless web browser session with Chrome... The Chrome engine, and then if that part doesn't work, it does the same thing, but then does a specific user agent to make it look like a human doing it. And then if that doesn't work, it actually boots up an interactive Chrome session. And then last but not least, an interactive Chrome session

that has to deal with one of those CAPTCHA dialogues that you often have to put in when you have to access something securely or if it's verifying your identity, it's got a method for that too. So yes, the author acknowledges here that's definitely a bit cumbersome, but chances are one of those four or five methods is going to work. So he does have an example here of doing this with both that stat can resource of data, as well as the city of Vancouver's

access for accessing the link from the API, and yep, it's got a shim function, which is called fetch. That's where you would use instead of like the typical download. File that you might do in VASAR. So once you... He shows what it... When it doesn't work, we'll just download. File, and then if you tried to do curl, if that didn't work directly, but then with safe conduct, it goes through these different methods, trying the headless chrome,

and then it has nice messages telling you what's happening behind the scenes in case you're curious. And then if it opens a window, and then if you have to click something in the window, you go ahead and click it. So there might be still some manual effort involved, but boy oh boy, I can see how that save a lot of time. You just want to get your hands on that data again and to be able to account for these different hoops that the provider is making you jump through. So there are interesting additional functions you can use in this, such as the probe function

that actually looks at what the server is returning as a response in case you need more diagnostics. If you had a chance to talk to their IT group or something, you might be able to pass that information to them. Not that I would pretend that you would get that opportunity because a lot of times these IT groups like to stay in the shadows. You heard it from me first. But nonetheless, it is a very intriguing package here, but that's not just the methods that he's provided here. He's got another package that takes it to another level with how you actually host and consume this data.

So Mike, it's got one of your favorite frameworks in this. Why don't you tell us all about it? Yeah, we're gonna talk a little bit about Parquet, but Eric, quick story time. Oh, I'm here for it. We have an open source R package out there called FCall, which downloads essentially financial statement filings that are public for, farm credit system banks in The United States. And that data was hosted on a website, just at URLs on this website,

in zip format that we unzipped and cleaned up. And we had a nice download data function that would hit the website and download and unzip the files for folks to then turn into tidy data frames and play around with. Well, they stuck a Cloudflare, layer on top of that website. So our, data download function did not work anymore. Oh, boy.... The Cloudflare layer. So this is, you know, a very timely blog post for me that happened fairly recently. And we ended up migrating

all of those files, kinda manually to an s three bucket. Uh-huh. So, introducing now the r two parquet package, getting back to the blog post here. The... There has been a a new package called r two parquet that has been developed, and it it all starts with hosting, these open datasets on Cloudflare r two, which is something similar to AWS s three buckets, which I was more familiar with. I was not familiar with Cloudflare r two, but it's essentially,

you know, an s three compatible service that allows you to host data and has some pretty good advantages, including sort of automatic global delivery and, some nice caching features as well. I think that may make the download, process a little bit more efficient than what you might find in other services. And, the choice here was to host the data in Parquet format. I certainly don't have to, tell you, Eric, about why Parquet is a fantastic format for storing data.

But for those that aren't initiated yet with Parquet, it's an open source column oriented data file format. It's very efficient. So if you have a large CSV file and you stored it as Parquet, the size of that file will probably decrease maybe in half, maybe more or less depending on really how big that file is in the first place. The column orientation allows you to just retrieve, you know, specific columns at a time instead of having to ingest the whole CSV.

And another benefit is that there's metadata stored in that file itself. Things like sums, data types as well are stored in the file. You don't have to read that file in. And as part of your data ingestion, set the data types and do all that parsing. It's already part of the file itself, is fantastic. But beyond Parquet, this R2 Parquet package has been developed to be able to ingest this data from, you know, for example, the the Cloudflare

R2 storage. There's also some spatial and metadata enhancements. There's automatic ordering of spatial data along a Hilbert curve, not something I'm familiar with kind of outside of my wheelhouse, but allowing for fast spatial queries using like at the ST filter function. There's reparation of invalid geometries. And it also embeds multilingual metadata. So like the title license and attribution directly into the file attributes itself. There's a couple of great examples of,

of code here and it it being used as well as, you know, some of the the the time savings that you're probably gonna get by ingesting some larger, Parquet files. There's a nice underscore metadata function, which is really nice to be able to read that that metadata alone without necessarily having to ingest the whole rest of the Parquet file. And some excellent sort of data viz, spatial data viz that's going on downstream with the results of some of these publicly hosted Parquet files. So

fantastic blog blog post. I'm excited about these new two packages because they absolutely do solve a lot of day to day problems that that I face. Yeah. I'm really intrigued by the different integrations that that we see here across the board. And another testament to if we can make these data easily accessible, even if we have to jump through hoops to get there, if other, the actual upstream providers aren't making it accessible, but then to take that processing,

put it in a performant format, send it to cloud storage, man alive, this is going to benefit a lot of people and a lot of the principles you can take here to whatever, you know, organization needs you may have. I see a lot of a lot of spatial emphasis here, which I know there have been big advancements in the Parquet and spatial space, but it looks like, you know, Jen here has has done a a great, great job of consolidating some of those updates here with acknowledging that you may also integrate this with some of the other external

packages out there, to get some additional functionality. But, yeah, you know, I had this misconception that Cloudflare was just for, like, making web pages faster with their CDN. Boy, folks, if you haven't followed Cloudflare anyway, this is not free advertising, I promise. They do have a lot of services. Although if they want a sponsor... Wait. No. I'm kidding. I'm kidding. I'm kidding. No one on their side will listen to this. I know a lot of my friends from the Linux circle are building tooling

that will integrate either their container registry service or the R2 service as a slightly cheaper, it seems like, method of object storage. So I'll have to play with this a bit, but it's great to see that from the R side of things, we can integrate nicely with this platform, but also just take a really annoying problem and make better front ends to it. So I think safe conduct can go hand in hand with other packages that are not so much trying to fix an access issue,

but are literally trying to be more polite about requests, and I say that intentionally because there is an R package called polite to help make your your request not, you know, do a a DDoS on a web server. So I feel like these two can go hand in hand in a lot of these newer ways of obtaining, online data and the like. It's that time again, Mike. We're gonna visit the, visualization corner for our next couple highlights,

and our, it goes without saying, boy, do I ever enjoy learning from one of the the key thought leaders in this space attacking a real applied problem that on the surface you may be wondering, why do we have to do that? I can tell you very specifically in a large organization, the method we're about to tell you about is very real and it is very valuable. What the heck am I alluding to here?

You know, it's one thing to make a great static chart in a graph, and we have many tools in R that can do that, and one of them will be the foundation of what this blog post provides, but when you want to have that interactivity feature without having to go the full blown shiny route in the case where maybe you don't have a way to host that up effectively, or the person that's going to be person or people consuming these visualization do not have the R expertise to launch an R script and run it on their own footing. There is a way to me in the middle here, and that is to be able to have an interactive chart

that you can provide via an email. Yes, this is important, and I'm very pleased to say that Nicola Reeney is tackling this head on in her latest blog post, the subject of our next highlight here. And, of course, she's been no stranger to telling us how to make a plot that you create with g g plot two interactive, but then it's not just that step, it's how do we share it. So what we'll frame the blog post as is that you start like you usually

do a visualization, you want to do a nice static representation of that data, so you can easily make a g g plot in this example, the diamonds dataset, get a nice scatter plot of the carat versus price, got a nice legend to facet, by the cut, and maybe that's not good enough. Maybe you want to give the user a way to interact with the plot itself. What if you want to add tooltips as you hover over things?

What if you want to add some additional features? And that is where a very underrated gem in the R ecosystem, g... Oh, oh, she's got a pronunciation guide. I'm gonna be perfect here. It is pronounced giraffe. Apparently, that comes straight from the author's source himself. Thank you, Nikola Nikola, for putting that reference in the blog post. That makes me as a podcaster's life a lot easier, doesn't it, Mike? That's been a long standing question, floating question out there for us. Hey, you know what? I'm

here for it. Thank you. Thank you. Now we will make sure we pronounce that correctly. And again, shout out to David Gohel. He is the author of Giraffe. It is an amazing package. So, in the blog post, she takes this static graph of the scatter plot and, first of all, just putting the foundation to make the plot interactive, where you get basically on the TIN a way to download that plot as a PNG or to make it full screen in your web browser.

Simple enough. But like I said, what if you want to add a tooltip? That is a simple mapping or an AES functional way with tooltip based on the variable you want to put in the tooltip, and in the blog post, you can literally see it in action right there. If you want to hover your cursor over these points, you'll get the cut of these points. Let's take it even farther. Right? Some I did not give a lot of, you know, a thought to would be possible, but it absolutely is. You can combine Giraffe

with some lower level in the R ecosystem of web development, input selector of sorts, not built with Shiny, but built with HTML tools, which Shiny uses under the hood to power a lot of, like, the select inputs, radio buttons, and the like, well, you can make a drop down menu with HTML tools quite easily with the use of the tags function and using the select function, which basically gives you that drop down,

feed in dynamically the names of the various options you want to put in there, which you can print out in your console to verify the HTML is correct. And yes, even in this blog post, we got the dropdown of the different cuts right there, right there, you can select through, but then you can link that to the giraffe powered ggplot, which will change the data based on that action or that selection of the choice by hooking it with a different action you get

with a little snippet of JavaScript, but she's got examples right here to make that happen. And again, I'm sure in today's day and age, you could get some easy help with your AI agent of choice to put that snippet in, and sure enough, in the blog post, I'm going through these different cuts and the the plot points are changing right on the fly. Absolutely perfect. So now you've got the plot, you've got interactive, you even got the little selector

widget on top of it. Now let's package it all up, and maybe package is wrong choice of words, you want a document format of this. Well, naturally, for doing of interactive content, it's going to be an HTML document, that's where the Quarto engine comes into play because you can basically embed the code to make this plot into a Quarto doc. The one thing to watch out for, in Quarto, by default, it will not put,

say, the underlying web dependencies in your document itself, it'll often put that as a separate directory that's being sourced when the user loads it. This has no issue when you're on, like, a deployed POSI Connect server to host portal docs or even a static site generator like GitHub Pages or GitLab Pages. But if we want to share this HTML file itself via email, it's got to be self contained. So there is an option in the HTML format

of embed resources of true. If you set that, then everything is in the HTML file that is encoded in, like, the headers and other CSS and other JavaScript bindings such that if I get this HTML file and I open it in my web browser as a local file, not going to, like, an online page, but going to a local file, it will just work. But if you don't do that embed option, you'll get a bunch of, like, missing errors and what not. Your

collab will be like, what the heck is this? They'll be mad at you. But with this, they won't be mad at you, and it gives you an easy way to have these interactive visuals just sent via the tried and true email when you don't have the luxury of a POSI Connect deployment or GitHub Pages to deploy to. This is a really quick way to quickly share it with somebody in an email, in a Teams chat, or whatever instant message client of choice so you can get on with your day.

Another practical tutorial that many in the industry will definitely resonate with across the board. Always something very practical to take away from a Nicole Orenny blog post, and this is certainly no exception. I love the walk through here. Love the use of the giraffe package as well and kind of this simple example demonstrating, you know, the ability to package up everything, as you mentioned, Derek, into a single HTML file that you can share, not just with the interactive chart,

but with also that little dropdown, which I think is, you know, may seem like something small, but that does introduce some, a fair amount of complexity as well. I think if you didn't have that that drop down, it would be a little bit more straightforward to just package up that interactive chart, alright, in the first place. Yeah. That embed resources, true flag, that is always a gotcha. If I can reemphasize that in this blog post, that'd be the one thing I'd reemphasize to make sure that when you do finally send that HTML file to somebody else, that it will actually work for them and and not just work on your machine because you have those additional files,

nearby in that same directory that are needed when you click on that HTML file. I don't know. You know, this seems a lot more complicated than just sharing my shiny app local host eighty eighty URL link to a colleague. But you know, to each their own. I'm just I'm just kidding but I have received a few. Emails saying, you know, hey, I got my app running over here. Local host, thirty eight thirty eight. Check it out. I I can't check it out. Sorry. That's just your computer. But this...

Not to rant. This this reminds me of of Quarto dashboards a bit as well. Yes. Though if you do use the shiny package in Quarto dashboards, you'll still need a non static server. Exactly. But one thing that I've played around with in the past and and loved and admittedly have gotten away from it is observable JS, Quarto has native support for. You can just add an OJS chunk. You can also pass objects between, you know, a Python or R chunk to OJS pretty easily.

And it's very similar to what Nicole is doing here. Maybe also avoids you needing to deal with the HTML tools package at all, involves learning a bit of OJS syntax, which is, again, very easy. And also, you know, allows you to have charts that have maybe select boxes that are, you know, dynamically render what's on the chart, drop downs, fairly simple filtering, but in a lot of cases, enough for sharing with somebody else. And again, you could use a Quarto document there with embed resources is true.

And I think accomplish kind of that same goal through OJS. So, again, you know, a few options out there. This is a fantastic one that Nicole runs us through. And I think, you know, something that I hadn't necessarily seen pointed out in the past. So really appreciate her taking the time to walk us through this. Yeah. Talking about you giving me flashbacks in my early, early days of Shiny and as I was, administering the Shiny server open source edition in the company company walls,

and I would tell people, hey, guess what? You built that great Shiny app. Stop sending those localhost URLs deployed on the Shiny server. That's how people can actually see it. It took a bit of time for people to learn that and, you know, that's a good been there. Yep. So, yeah, good good memories on that one. But, yeah, it it is interesting. There are a lot of movements to go, quote unquote, serverless with a lot of web resources, so it's great to see that we got a bunch of options here

that really take advantage of these humble little web browsers that most of us take for granted as just, hey. I'm going to Google to search or something, or I'm gonna check my email or something. They really are like many computers under the hood. When you start diving into, like, your web tools, developer console, you'll see all sorts of things happening on what we call the client side that really are hugely powerful. Of course, there will be people that say that JavaScript is just like everywhere too much, but at the same time, you can't discount the fact that when you get capabilities like this, it is quite powerful, and I think there's a use case for everything. So even with the observable JS,

you know, integration with Quartal that you mentioned, for my other hat that I wear with the submissions working group, I have in my Quartal site for that working group itself a meeting minutes kind of template generator where I do have an observable JS block to choose from a list of attendees and update a draft of the courtal doc on the fly. I, that was super geeky. I'm the only one that really uses it, but it was a good excuse to learn what you can do with observable JS. So I,

greatly appreciate playing with that tech. And yes, if that scares you a bit, don't worry, you got an agent to help you with a lot of that stuff, and it will definitely, you know, give you a lot of lens into what's powerful. So, again, as always, Nikola does a tremendous job with the blog posts, all the examples, and everything in between.

And rounding out our highlights today, we're gonna stay on the visualization corner of sorts, and we're gonna give a nice set of kudos and some thoughts on a awesome workshop that was recently, shared with the main R user group. Great state. Great state. Absolutely. And And this workshop that was held virtually was led by another key thought leader in the world of visualization,

Kara Thompson, who runs her own consulting for visualization. And this workshop was held, I believe, in October, which tackles a lot of the planning and the principles that you should have as you're thinking about not just building that visualization, but communicating that with your intended audience. Full disclosure, at the time of this recording, when this blog post was shared, I do not have access to the actual recording of the workshop, so I am going by what I've seen on the slides that have been embedded in this post along with the resources that have been shared in

the tail end of the blog post. But some principles that I see here that I think in the world of visualization are a must when you are building these for not just yourself, are really thinking about what questions are going to be prudent to the end user of these visualizations, and sometimes it's hard to know what those what those should be.

So it's one thing to jump into, hey. You got your data and then here. If you're a fan of penguins, you'll love these workshop slides because it's the Palmer penguins dataset all the time with, with, awesome GIFs of penguins. So, awesome job, Kara. Love the visuals there. But as you look at your data, think about, for someone new to that data, what kind of questions they might ask. She's got a few examples here of, like, what island is most prevalent with the penguins, you know, populating them.

Are there weird features of their bodies such as, like, maybe their beak dimension? What about, you know, certain types of penguins? How prevalent are they? And you use that to figure out, okay, how do those questions map to different aesthetics in your plot? What should you use as legend items? What should you differentiate by colors? Maybe there's things you have to do for the axes, where regions you're zooming in on.

What is the size of the points you want to consider? And throughout it, and much like the technique that was pioneered years and years ago, like that g g plot flipbook kind of paradigm. In the slides, as you navigate through the examples, she highlights different portions of the code to see what impact that has on the plot in a layer by layer fashion. GGPlot2, when you build plots, it is very much like layering a cake, if you will, or a recipe.

You build and build and build until you get to the point you need. So it's all about finding the right colors that are tailored to maybe the data at hand, but not just, you know, finding some fun color palettes, checking that they meet accessibility requirements because you want to make this as visually consumable for all audiences. So she's got links to different, resources to check the differentiation

between colors as well as making sure that they are accessible to different perspectives, such as like from the monochrome perspective. Definitely check those links out at the end of the post here. But yeah, it's a great mix of both scatter plots, kind of like hybrid box plot, density plots, ways you can tailor the titles, the subtitles to meet the needs of it, and really having a cohesive brand, a cohesive,

you might say, paradigm for how everything looks when put together as you go from one visual to another, feeling like you're still in the same experience. So there is a lot of code that you can you can choose to play with on your own, but I love the layered approach. I love the principles of starting with those questions and then really figuring out which aesthetics tailor to those needs.

So, I can't wait for the recording to go out there. I will put that in my media server to watch that next time I'm getting ready for a visualization task. The slides are excellent. And as I said, it's not just about what you think a visualization is appealing as, it's what do you, what is the intended audience and what are the key questions that you should be asking and answering with the features that come standard in ggPod two and say the vast extension ecosystem on top of that? Yeah. Absolutely.

You know, I see so many best practices here in the slides and in full code snippet that is at the bottom of this blog post, which is I think the final result that Kara puts together. One of my favorite things that she does is she spends a lot of time on theme, and so much so that she actually extracts the ggplot theme that she uses into its own function, which I I really love because then when you look at the rest of the ggplot code, it's pretty easy to digest

as opposed to having to look at, you know, a very, very long set of code where you're kind of identifying right columns and different variables that you're passing, intercepts and things like that, and mixing that in with the actual theming elements, is kind of the opposite side of the brain. And when you take a look at the theme, it might look like a lot. But this is really the thing that makes this ggplot plot not look like ggplot. And Eric, you and I talk about all the time trying to make shiny apps that don't look like shiny apps. And this is a fantastic case of making a gg plot that doesn't look like a GG plot. There's so much to learn here even just from this one plot that she puts together. I mean, we have a a little annotation up in the top right hand corner that's perfectly placed. It's got an emoji in it and it it looks like, you know, somebody made a meme

you know, with the snipping tool or or Photoshop or something like that. But no, it's it's all g g plot all the time. It's fantastic. The the way that the g g text package, think, is used to do a lot of this annotation and handling this text and just making it making it really, really pretty. So a a lot to learn here, from the slides, the blog post, and hopefully, eventually, as you mentioned, Eric, that recording. Yeah. It... It's amazing, amazing how far the ecosystem has come in the world of visualization,

which, as you've got to, is that foundational layer. And it's great to see, I'm glad you called it the theme thing because I remember wrestling that head on in a day job project years and years ago that I affectionately, maybe not so affectionately, coined as the glorified PowerPoint maker, where I did need to adhere to a very, and I do mean a very rigid brand guideline for that given asset that given, you know, that they use for communications.

It was painful. But once I figured out the fact that, hey, with these themes, I can either, a, take an existing theme, maybe add a little bit, or just make my own theme completely custom. But not just make it custom and then copy paste the darn thing in every g g p two call I make, but to make it one utility function with the default set as they are and only have to change maybe one or two for a given type, that code was so much easier to review and to collaborate on with the key SMEs that were consuming these slides

and making sure that it was on point with the brand. If I had to change it, I only had to change it in one place instead of like around five or six different plots or maybe 50 or 60 plots, depending on how far they took my humble little app. So the principles that you learn, even if you feel like, oh, in visualization, it's very much kind of a creative process, and, yeah, it is creative, but you can still have good programming practices to make it as easily maintainable and as easy to collaborate with as possible.

So I think it's great that ggplot2, in its vast array of ecosystem, still has ways for you to follow those principles and, in this case, have a really polished plot afterwards that if I didn't know this was an r based post and I only saw that end product, I would not be able to tell the CGI plot to. I simply wouldn't. It is that good. Absolutely. No. I agree. And I do remember a lot of pre show conversations about your PowerPoint generator app, Eric. Fortunately,

I think it has been softly retired. Hopefully, the people that, you know, we'll leave it at that. Because I like making slides in different formats, folks, and I think you know what ones they are because they are web based formats. That is my passion, just like everything on the web now. Right? And what else is on the web? Well, it's not like you get rweekly in the mail, although that would be kind of fun. You get it as a web page you can go to every single week whenever you want, however you want, mobile, desktop, it doesn't matter. We got a great interface for you to look at. All of the additional resources in this issue

on top of the highlights that we have here. Lots of great tidbits here. I see even, you know, on time to cover all of it today, but a nice roundup of some hidden gem of r functions that you may take for granted both in base R as well as the the very vast R ecosystem. And even for all of you, basketball fans out there, some great looking at how FiveThirtyEight is, doing some of their NBA probability visualization and summaries. So cool stuff there for the basketball fans now that NBA is back up and running.

But, we got a lot more in the issue as well as always, and also where do you find it? It is at rweekly.org. That is where you go for everything related to rweekly, including how to do a poll request. Yes. Yes. I have not made that friendly app yet for a poll request, but I am gonna speak it into existence until I get there. But until then, you can just hit the little octo can in the upper right corner, and that is where if you want to share that great new resource, that great new blog post, whether from you or someone else that you found,

we are just a poll request away. So hit that and you will be sure to get that into into the, into the issue as well. And, also, we love to get in touch with all of you, and if you have feedback about the show, we have a feedback form. Did not get any email or feedback about the Mr. Unhinged over here, so I still have a podcasting job apparently. But if you do have other messages you want to send us our way, go to that contact page and you'll... In the show notes, and you'll be able to send that to us as well as getting in touch with us on social media, which is

for me it's bluesky where I'm rpodcastbsky.social. I'm on Mastodon where I'm rpodcastspodcastindex.social, and I'm on LinkedIn. Search my name and you'll find me there. And Mike, before we get to your social handles, I know you wanna say a very important, word out there before we close out here. Yeah. I lost a really close friend who was a member of the R community as well, Brad Lindblad, this this past week after his second battle

with cancer and I Brad was the author of some great R packages out there. Probably most famously the the Shrew package which he published for R, Python, and Julia and I think it was used in Tidy Tuesday and and some other things for text mining. It contains the whole transcript from the office and a lot of my and relationship revolved around the intersection of programming and and the office which probably lends itself to to his personality which was fantastic and he was a big mentor of mine. Really taught me everything that I ever knew about, you know, software development, best practices, and,

getting me over from Basar to the Tidyverse in the in the early days of my R education, and he was a mentor to to a lot of other people as well. So, if you feel so inclined, there's a couple ways to support his family as they they navigate his loss that will put in the show notes but thanks for giving me a minute to to share that there but if you want to get in touch with me personally on social media, you can find me on LinkedIn. Probably best if you search Catchbrooke Analytics, k e t c h b r o o k. You can see what I'm up to. Or on blue sky, mike dash thomas dot b s k y dot social.

Awesome. Thank you so much, Mike, and, yeah, our sincere condolences to Brad's family. It is always, difficult when you you lose some of your heroes, your mentors, and those that have impacted the community greatly. Yeah, I definitely sympathize with that. So sending all the good vibes and the best wishes to Brad's family and also in the general world. We also lost a legendary, actor that a lot of us on the geeky side had high value with.

Make it so, Patrick Stewart. We miss you, buddy. Yeah. Well, with that, we hope that you have a wonderful weekend. We, again, thank you so much for listening to episode 247 of our weekly highlights, and we'll be back with another edition of our weekly highlights next week.

Open the transcript file

Hosts and Guests

Hosts

  • Eric Nantz Host Feed-stated
  • Mike Thomas Host Feed-stated

Credit the people on this episode

Be the first to rate this episode

More episodes

All episodes
  1. Issue 2026-W47 Highlights

    Nov 20, 202648 minAudio

    Our takes on the increasing removal of friction in good and not-so-good ways thanks to AI, a visual tour of hot air balloon accident data, and a case study of novel image recognition with the rplatipus package. Episode...

    Credit the people on this episode

  2. Issue 2026-W46 Highlights

    Nov 13, 202633 minAudio

    The flow of both data and analytical jobs just got smoother with mizu, opening the door for multiple agent harnesses in Positron, and an optimal statistical model tailored to complex experimental designs with a friendly...

    Credit the people on this episode

  3. Issue 2026-W44 Highlights

    Oct 30, 202631 minAudio

    The celebration of rOpenSci's 15-year anniversary continues with the community sharing their perspectives, and a fun Shiny app that could be quite useful during celebrations on a weekend or the holidays. If you ever...

    Credit the people on this episode

  4. Issue 2026-W43 Highlights

    Oct 23, 202659 minAudio

    We learn about the variety of ways you can contribute back to the core of R, a roundup of the excellent choices available for package and environment reproducibility, and an amazing collection of Shiny applications and...

    Credit the people on this episode

  5. Issue 2026-W41 Highlights

    Oct 9, 202652 minAudio

    A roundtrip between formatted tables and spreadsheets is now very much possible, how {shinyreact} represents a new path for user interface development in Shiny, and a candid take on ownership of code and the products we...

    Credit the people on this episode