Allison Horst is an artist and a data scientist advocate at Observable. She’s created a library of open artwork for data science and statistics education and has also been a painter her entire life, focusing on landscape and plein air oil painting. You can find her work at https://allisonhorst.com/.

Can you start by describing the art you make?

My artwork mostly falls into two categories that are very different: fine art landscapes, and small digital monsters who love data science. The niche that I am in for illustration, as related to data science, is cartoons that highlight features, functions, or methods, using little friendly characters to try to make the content more welcoming for learners and users, especially in the R community.

When did you start thinking about data in relation to art?

The first time I thought I could do illustrations that would help me understand or communicate something about science was in undergrad and grad school. I started doing it consistently when I started teaching stats in 2012 and looked around at the documentation I was sending to students and thought, “This is boring and intimidating.” It was certainly not inspiring. I started thinking that I could draw little designs to make this feel less sterile of a field—a more fun field, a more fun class. So, in 2012 and 2013, I started making a bunch of illustrations for my stats class.

At that point I was drawing with pen and ink on paper, which is awful, because then you take a picture with your Coolpix camera with terrible white balance, and then you get to your computer and you crop the photo and try to make it look good, but it’s pretty horrible no matter what you do. So it was not a fun process of trying to transfer what I had drawn to my course materials, but I did that for five years. The game changer was when I got an iPad with Procreate.

[Working digitally] was truly a life-changing experience, like oh my goodness, I can draw this thing, and I can just share it immediately! I don’t have to fuss with all the parts of sharing artwork that I’m not interested in; it’s just immediate. And I can press Undo, which I think was the most empowering thing I have ever felt as an artist—being able to undo something. What people, I think, now recognize as my data science artwork was what I started when I began working digitally.

It’s amazing how technology can change things. The level of risk-taking you’re allowed to take in being creative with a tool like the iPad is incredible. If I had been using pen and paper or paint and paper with these little monsters and I wondered, “What happens if I put a hat on one of them?” the risk is that if it doesn’t work, then I’m throwing the whole thing in the trash. Pressing Undo sounds like such a simple thing, but what it really means is that you can take risks that you otherwise wouldn’t have. And that’s a pretty amazing thing for an artist.

What is your process like? How do you go from idea to project/product?

It’s just mostly chaos. It usually starts with a function I haven’t used before or wanting to understand something better, or sometimes I just have an idea in my head that I have to draw. Sometimes it’s just an idea that makes me happy—for example, my Halloween illustrations. They don’t teach anybody anything, but they make me happy. Other times, I think it would help bring awareness to a cool tool or make this function feel a little bit more friendly, especially for people who haven’t used it before. But I don’t have a list, like, “I’m going to do this and this and this and this”—and that’s really great, because it’s not a job. It’s just a thing that I love to do when I feel inspired, so that keeps it really fun.

When I get one of those ideas, usually nothing else is happening until this thing is done. I think a lot of artists work that way. I’m not very good at dedicating a half hour each day to a piece of artwork, because I have other projects to do. For me, if I have an art idea, often that means I’m staying up until three thirty this morning because I can’t focus on anything else until it’s done.

Another thing that working digitally has done is cut down the time I spend on planning artwork; I can just start and see where it goes. I can always move stuff around right on the page. If I think things are too close together or not really portraying the relationship between these two monsters that I want, then I can move them. So, I think I’ve spent a lot less time planning and sketching out ideas, because I mostly just go for it and see what happens. Sometimes it ends up being something that’s worth sharing.

How does your background in plein air landscape oil painting influence your current work?

I’ve been painting since I was little, since before I can remember. I started oil painting when I was in junior high. My parents got me an oil painting set and I was so excited. I went upstairs and I painted what I thought was this beautiful sunset. I came downstairs, and they were so encouraging. They told me it was awesome. They still have that painting. I look at that first painting now, and it looks like there is a nuclear bomb going off over a terrible ocean! I remember my grandpa was there, and I remember him looking at one corner of the painting and saying, “I think these colors are really successful,” and at the time, I remember feeling really proud. Now, I realize how forgiving and encouraging they were.

After that, I just painted all the time—all through high school, all through college. I spent quite a bit of time showing my artwork at galleries around Santa Barbara through college and grad school. Most of what I do is landscape paintings. My favorite is plein air landscape painting, which is painting outside on-site, usually smaller scale in pretty short amounts of time, to try to capture the light before it moves on. I still do it, and I still really love landscape painting both on canvas and oils and also digitally.

The way that I think about both landscape painting and the little data science monsters (and really any illustration that I do) is that all of it is trying to engage a viewer in a way that makes them want to learn more about something, in a way that makes them want to walk into it—whether that’s an R function or a mountain landscape.

With landscape painting, you usually don’t want a straight hedge across the front of a canvas because that doesn’t give a viewer a way in. So in a lot of paintings, the foreground will have either a literal pathway into it, or a visual path that guides you into the painting. Often the lines of the painting (or the landscape) will pull you back in. With the monster illustrations, I try to do the same thing. Like, hey, come hang out for a while—you might learn something, see something in a different way, or feel inspired to investigate something a bit more.

Even in these different genres of art—plein air landscape painting and digital monsters doing data science—though they’re stylistically totally different and the media is totally different, they’re both trying to share an entryway into something in a way that engages and resonates with people.

Are there environments in which you feel most inspired?

I always feel inspired when I’m in the mountains, and in the high desert. We live up in the mountains in a place where we don’t see a lot of people—most days we see more coyotes than humans. Being out in nature and in wild places is usually where I’m happiest and most inspired to make artwork.

What is the space between art and data? How does your work fit into this space?

The space between art and data is a space where we’re asking creative questions and trying to be successful in communicating our answers. Whether you’re a data scientist or an artist, the commonality is that you’re inspired to try and learn something and share what you’ve learned with other people. With art, it’s learning about the light on this mountain or the shadow, or how to paint a rock. You’re trying to learn something about the experience and a sense of place and thinking about how to share this feeling, because maybe it resonates with some other people. As a data scientist, we, too, have questions, and we have other tools (like code instead of a paintbrush) to try to help us explore and answer them. And then we have places like canvases or notebooks to try to share what we’ve learned in the creative process.

Art and data science are both totally exhausting. I think they’re similar in those ways, debugging a painting and debugging code. If you try to do it when you’re tired, you’re going to overwork your painting and make a mud puddle. And it’s the same thing if I try to debug code when I’m tired—it’s just nonproductive frustration and rage.

It’s easy to think that data science is sterile and noncreative, like there’s only one way to write code, but there are as many styles, languages, flavors, personal preferences, and opinions of writing code and solving problems with data as there are for illustration. There’s so much creativity in the way people can work with data, and I think that far too often when people say “creativity” and “data science” together, they’re only referring to data visualization. That ignores the whole creative process of working with data, which is very mind-twisting—rethinking and reenvisioning, finding creative solutions, searching for the right tools, throwing stuff away. It’s an imaginative process, and I think that’s why it’s so hard to teach data science. It does feel like as much of an art as a science: everything is a gray area, all of the colors overlap, and you have fifty different paintbrushes you could consider using. But I think that perception of math and stats as a flow chart, where you just pick and make decisions at each node, doesn’t really reflect the creativity that goes on, and it misses the nuanced thinking and practices in data science.

What impact do you hope your work has on the world?

I hope it makes people feel a little bit more welcome when they start learning R and data science. I hope it makes at least a few people think, “Oh, this isn’t just a place for people who already know how to read code documentation—maybe this is a place where there’s extra effort put into making it feel friendly, so I’ll give it a shot.” So that’s one hope, that more people feel more welcome when they start learning. The second is I hope it just brings some joy to anybody at any level. I hope it brings a little bit of levity. A lot of folks working in data science are working on really challenging and huge pressing questions, so I hope it brings them some brief happiness. And third, I hope that it maybe helps some people learn and remember some useful things about data science. But that’s definitely third place—the welcoming and joy are bigger priorities for me, and if some better understanding or engagement or willingness to look up a function and try it out comes from those things, then that’s even better.

When I first started, I felt the pressure that my data science artwork needed to be didactic and have clear learning objectives. But I’m really happy about the shift to not feeling pressure to have learning objectives for everything. That way, I can focus on just making people feel a little bit better as they’re learning hard things—that is my top priority with the art.

Is there anyone else’s artwork that you are in love with lately?

I recently saw Helga Stenzel’s work on Twitter and almost lost my mind, and I’m still trying to resist buying every print on her site (https://www.helgastentzel.com/). I love the Laundrosaurus so much. I think that her transformation of very normal, everyday things into artwork is just really interesting and funny. I feel like there’s so much comedy in it. Her ideas are just brilliant.

This interview has been shortened and edited for clarity and readability.

Interviewer’s Bio:

Jenn Schilling is a senior research analyst at the University of Arizona, a data science mentor at Posit, and founder of Schilling Data Studio, a data visualization training and consulting agency. Jenn has over a decade of experience working with data in a variety of industries, and through Schilling Data Studio she helps analysts and aspiring analysts grow their data viz, analysis, and R skills so they can effectively communicate data and insights.