Dr. Matthews is Associate Professor of Statistics and Director of the Center for Data Science and Consulting at Loyola University. He also is a data artist who developed and promoted the Data Art Show, which debuted at the 2016 Joint Statistical Meetings. He performs with the Uncontrolled Variables comedy troupe at the Lincoln Lodge in Chicago and you can see his data art, links to his comedy performance, and much more at his website, Stats in the Wild.
Episode Description
A statistician walks into a bar, and a comedy and art show begins. Creative work for scholars can extend beyond novel research and application. In today's episode of stats and stories, we see how the intersection between interest in statistics and art, as well as the intersection of statistics and comedy, with Dr Greg Matthews.
+Full Transcript
John Bailer A statistician walks into a bar and a comedy and art show begins. Creative work for scholars can extend outside of novel research and application. In today's episode of stats and stories, we see how the intersection between interest in statistics and art, as well as the intersection of statistics and comedy, is realized. I'm John Bailer. Stats and stories is a production of the American Statistical Association as well as Miami University's departments of statistics and media, journalism and film. I'm joined in the studio by Rosemary Pennington from the Department of media, journalism and film. Our guest today is Dr Greg Matthews, Associate Professor of statistics and Director of the Center for data science and consulting at Loyola University. He also is a data artist who developed and promoted the data art show that first appeared at the ASA joint statistical meetings in 2016 and he performs with the uncontrolled variables comedy troupe. You can see his data art, links to his comedy performance and much more at his website, stats in the wild.com,
John Bailer
Greg, thank you so much for being with us today.
Greg Matthews
Oh, it's great to be here. I love talking about statistics and art and comedy.
John Bailer
Well, thank goodness, because we weren't prepared for anything else. Perfect. I have to thank you for giving so many, so many options of where and how to begin this episode. You know, I was just spinning my wheels for a while, but I guess I think I'll start with art. So, so how about just the a basic idea of how you would differentiate data art from, say, data visualization,
Greg Matthews
that's a spectacular question. So I have this sort of like corny answer for the difference between data visualization and data art. I give a I give a data art talk, and this is what I always say, is the difference data visualization answers the question. Data art asks a question, pretty deep, right? And the idea is that data visualization, I mean, there's certainly overlap, but in my mind, data visualization is essentially functional. You're trying to answer a question, you're trying to summarize data, you're trying to convey information to a viewer. Whereas data visualizations can become art when presented in the in the in the correct context. But they don't always have to be good data visualizations, right? The goals are different data like, you know, like a pie charts, not a very good data visualization. And there's rules to good data visualization, just like there's rules to good art, and you can sort of break those rules of data visualization when you're making data art. But I do see a lot of overlap in these two these two ideas, visualization and an art, and I think depending on the context, they could be one in the same, depending on how you presented it to a viewer.
Rosemary Pennington
I was looking around on your website at some of your art, and it struck me that if I did not know that you were doing this based off of data, I would have had no idea, right? It just looks like really interesting abstract art sometimes. How did you get started creating this kind of art? Well, so
Greg Matthews
it's my wife. My wife gets all the credit. So my wife went to art school, and we started dating, we got married, and over the course of, you know, our our our time together. I basically got in an art school education, and she sort of got me really interested in art. And then I started thinking about, Oh, I can, I can make art with my computer, and I can, I can make data art. And so that's what got me hooked. And then I just started doing it. And I really, I really like doing it. And what. And to your other point about how you don't you didn't know that it was data art. My goal, when I make data art is to make something that will stand on its own, and you can look at it visually, and it's interesting by itself, and then when you learn that there's data behind it, it becomes even more interesting. That's That's my goal. That's like my what I view as success in something that I've created artistically. It doesn't always work, but, like, that's my goal when I'm making it, it's more interesting when you when you realize there's data
John Bailer
behind it. Oh, that I agree. And, you know, that's, that's really cool, just to see this process. Can, can you talk a little bit about the process that you follow when you're producing a piece of data art? You know, I looked at, you know, looking through some of your collections that you have online, there's, there's some analysis and rendering that seems like it's that's hiding behind the scenes here. And so could you talk a
Greg Matthews
little bit about that? Do you know which pieces you're specifically talking about? So how
John Bailer
about, how about, no, just all of them. We have about five minutes go through it. Ready Start. How about, how about celebrity and gun. Okay, so a couple of recent ones.
Greg Matthews
Yeah, so these are from the Google Image Search series. So what I've been trying to do more recently is, you know, last five years, or whatever, five or six years, I'm trying to find data sets that are, like, more interesting, and I'm interested in exploring, like, data sets that are personal. And so I've done work with, like my Fitbit data and and things like that. But with the Google image search data, the way these are created is I'm doing a Google image search, and I'm saying I pick a word, so I the word was gun, or what was the other one you mentioned? Celebrity, celebrity. So I google celebrity, or gun, or any of these other words, and I and Google will give you back a set of images. And these images that it gives you back are it's what Google thinks you think those words mean at that particular time, right? If I Google the word, if you Google that word at a different time, you're going to get different results than I'll get. So there's something very personal about what is returned from Google. It's about the time and the place and your personal search history that you're getting these results back. So it's like, what it's what Google thinks I think this word means at that time. So I'm taking these images and I'm trying to put them into a composite image. Now there's certain ways you could do this. You could take the pixels and you could, like, average them, but this gives you kind of what I think are kind of either not boring images, but they're not as exciting visually as they could be. And there's actually an artist at University of Chicago who did this 20 years ago, taking images and making composites. His name is Jason saliva, and he has a lot of really interesting work. So what I was trying to do is make these composites. They're actually made using cart models. Where I'm I'm building a cart model to try to predict the color of each pixel. And now you could do this very well. You could build a very accurate cart model to predict the color of the pixel, the average color of the pixel, very accurately. But it's not that interesting visually. So what I'm doing is I'm actually I want to produce a cart model that is doing worse than you could, because it creates more interesting visual images. And so by creating a bad statistical model, I think you get more interesting visual images. And there's this idea in art where you need to understand the rules before you can break them, right? So I think there's analogies here where I have to understand what a good cart model looks like so that I can break the rules to make art. And that kind of connection really brings me a little bit of joy.
Rosemary Pennington
I wonder what your process is like and how you developed it, right? Because I, as you know, John pointed out, there's gotta be layers and of this, and there's lots of analysis going behind on behind the scenes that we don't see. But when you are trying to figure out what you want to create, like, how are you identifying the data, the process that you're going to use? And how do you like, what is your process for for getting to an image that you are happy with
Greg Matthews
well, so I'll usually start by picking a data set that I think is interesting, and so like I had the Google, the Google, Google Images, or fit it, I'm looking around my office. But what else I was using? I have some some work done that was based on the GPS coordinates your phone is tracking you constantly, right? So I have these images that are like the, I know it's a podcast, you can't see any of this stuff, but they're, they're, they're just like, single days of my life based on where my phone tracked me, right? And that's that's really interesting personal data and and I think good art asks a question, and one of the questions that I want to ask is about, you know, are we okay with all this data being collected about us? Right? There's just constant data being collected. And so if I find a data set I think is interesting, I'll start with that, then I just start writing code in R. So I do this, I do all this stuff in R, and. I will just go through, I'll just start with, like, you know, basic data visualizations. I'll start plotting pixels, and I'll see, all right, what if I write code like this? What does that look like? And then it's just, you know, you try 1000 things, you find the one that you like, and then that's what, that's what you you see, right? It's very much like doing statistical research. I you fail 10,000 times, and then what succeeds you write in a paper. And so people only see the successes, but there's a, there's a lot of trial and error, or, like, you know, you try something and you think it's you get excited about it, and then you look at the image, and you go, I don't really like that. So it's just a, it's just a lot of, you know, experimentation, trying something and seeing if it works, and seeing if it works and if it doesn't work,
Rosemary Pennington
you try again. You said you got into art via your wife. What does your wife think of your data art?
Greg Matthews
She thinks I'm a paradigm shifting, generational genius. She likes it. She thinks it's really good. She will she we have a relationship where, like, I can make something and she can, she can genuinely say that's good or that's bad. Here's why I don't like it. And so one of the things she's taught me is, when she was in art school, they would do like these critiques, and you had to say, I like something or I don't like something, and then you had to explain why. And that process her doing that with me, and then her getting me to do that has been really helpful in thinking about art in a different way for me, you know, I think a lot of people think about art as, you know, paintings from the 1500s about, you know, they're depicting, like, biblical scenes, like, that's what art is for a lot of people, like in a museum. But like, art is so much broader than that, and I had no idea about this until, you know, well, I met my wife, and we started talking about art a lot, right? But like, she will give me feedback, and she'll say, I like this, or this is working. This is what's good about it, and this is what's bad about it. And you could try to do something better here. So, like, we have that kind of, she helps you with that kind of stuff.
John Bailer
That's, yeah, I look you through some of your pictures. I mean, some of them, like, like beach, for example, seemed like it was really cut from kind of this impressionist cloth. Or, like the, you know, sometimes I've seen, like, the Scottish colorist cloth that that has that, that lovely image, those lovely colors. So you seem like you have this this recently, a lot of times these prompt from from image searches that are there. How has your art, your data art, changed over time? Do you find, where did you start, and what are some of the paths towards where you've where you've come to now?
Greg Matthews
So when I start, so when I started making art, I would describe, I make a distinction between computer art and data art. And when I started, I was making a lot more computer art. And what I mean by that is I was generating things randomly. So some of the other I did a lot of games of chance. So I was messing around with things like dominoes, Powerball, Craps. I have keynote in here, and I was using those as, like, a process of generating images randomly, using, like, you know, ours pseudo random number generator to generate images, and then picking the ones, picking the random numbers that I liked. But I don't that's not really data are because it's not, there's no data behind it. I'm just generating the images. And so at some point I did this for I did this for years at the beginning, and at some point I said, you know, I would, I would like there to be something more behind this than completely randomly generated numbers, even though I think a lot of this stuff is very visually interesting. But as I said before, once you find out there's data behind it, it adds a layer to it. And so I started shifting towards using using data as the the primary source of the art, right? And I like to pick data sets that have, you know, some kind of meaning beyond just, you know, here's a random data set from Kaggle, right? So I like to use a lot of my I like to use a lot of data that's generated by myself, because I think that gets, if other people are seeing it there, they can potentially think about what data is being collected about me. Are we okay with that? These are, like, big questions that I don't think people think about at all. They just use their phone all the time, and we don't. I mean, if you really think about it, you know that there's a ton being collected on you, but I think we just don't think about it's just part of life every day, right? But I've also, I've also, I've also worked with data that's like the census data about wealth and race, right? I had a series called American money, and it looked at difference. It looked at how money is distributed by within zip codes, and then the demographic makeup of those zip codes. And you know, you can see big differences between these things and but that it also stands alone as interesting images. But then when you see what's behind it, it's even more interesting. So I like to pick data sets that are, you know, meaningful in some way, and they sort of ask bigger questions. It's not always successful, but like, that's what making art is. It's trial and error. Mm. Yeah.
John Bailer
So you're listening to stats and stories, and we're talking to Greg Matthews about data art and comedy. I think we're, it's about time for us to shift to maybe some uncontrolled variables. All right, spectacular. So how did you get involved? What is uncontrolled variables? And how did you first get involved?
Greg Matthews
So uncontrolled variables is a science and comedy show. And the way it works is we, we bring, we get some Chicago area comedians, and we get Chicago area scientists, and we do a show together. So the basic premise of the show is a comedian comes on, and they do a regular set of comedy. And then we, we give them the we give them the scientist slides, and they present the slides, never having seen those slides before. And hilarity ensues as as you know. And then we bring the real scientist up, and they, they present the slides again, and they, like, you know, address all the things that the comedian screwed up, which is everything, and so we get so the audience gets comedy, and they're actually getting to see a real scientist, and they get to talk about the work that they're doing. My involvement in the show is I help produce it, I help find the scientists, but I also do a, what we call a guest lecture, and I take the topic of the show, and I do a data and I do a, like, a completely absurd data analysis related to the topic of the show. We actually had a show last night. We do it once a month, and you just happen to, we just happen to be talking the day after the show. Last night's theme was environmental science. So I did a data analysis on enteric fermentation, which is livestock flatulence, and about how methane is released into the air by livestock. And I killed,
Rosemary Pennington
you know, it's we talk a lot on stats and stories about how, you know, we're living in this environment where there does seem to be a bit of distrust around science and expertise and facts. And I wonder, as you are working to produce these shows, how you are, who you're imagining your audience is for this, and sort of how you're imagining this, interfacing with this larger, sort of cultural distrust of science.
Greg Matthews
So, I mean, I know exactly who the audience is. The the audience is a lot of graduate students in, you know, science in STEM and there's a lot of professors who show up. We sort of have a whatever scientist we bring in, they'll bring all their colleagues, they'll bring the people in their labs. So it's sort of like a comedy show cheat code, where we are guaranteed at least, you know, 15 people to show up, because the scientists will bring a bunch of people, because it's their one, their one time to be on stage. That's, that's sort of who the audience is, like, we're not reaching a huge it's not like random Chicago. And generally, though, there are people who just show up and they'll, they'll go, I want to go see a show tonight. And they end up at the show not really knowing what it is. And they seem to, they seem to like it. Your other question, though, about the culture, I think is really interesting. So I hadn't thought about this for like, ever. Like, okay, so there's, there's a woman who is currently filming a documentary about our show, and we did an interview. I did an interview with her for the documentary, I don't know, three weeks ago, and she was asked, she asked sort of the same question. She goes, What do you think about like when you do this show? Is there a goal of trying to reach a bigger audience, or, like, is there like a serious goal of trying to communicate science to the general population? And when I started, when I when I first got involved the show, I didn't start the show when I first got involved in the show, it was no it's just a fun science and comedy show. But since, I don't know if you know what happened in last November, but since then, the show is sort of taken on, I do feel like there's a little bit more, there's a little bit more of a serious side to that. There's a little bit of resistance in doing the show, right? So we just did a show in environmental science, and we talked about global we talked about greenhouse gasses and global warming, and I think the fact that we're just talking about this in this environment, it's a little bit resistancey. It feels different than it did, you know, a year ago or two years ago, and we're like, not we're leaning into this. We're going to do a show on. On the biology of gender in June. Oh, right, yeah, we're bringing in, we're bringing in a biologist who is going to talk about, you know, gender from a bio he studies lizards. He's going to talk about the biology of lizards. But like doing those shows a year ago, would be very different than doing those shows now. And I hadn't really thought about this, but I do think there's something a little bit more serious now about even just talking about this stuff, right? And so I feel a different sort of I don't know if responsibility is the right word, but it does feel different to me in a way. Does that make sense? Yeah,
John Bailer
absolutely. So as you think about next month and preparing for this, you know, you're going to be giving a guest lecture again, yes. So, so what's, what is the, when you start thinking about the, you know, the topic going in, how do you start kind of finding that, that hook, that connection, that you want to really, really build on, I mean, the Cal flatulence, that seems like a really good call for environmental science piece. So, so where do you Where are you going to start now in this process for the next month?
Greg Matthews
So this the the process is, I panic for a week trying to figure out what I'm going to do. When I settle on that. Then I do some data analysis. I see what the results are. I go, that's funny, that's funny, that's funny. And then I put them in slides, and then I like, add jokes in at the very, very end, right? If you can get a decent analysis, the jokes like write themselves. You just find some like, everything's funny about it's easy to make fun of science. I think the hard part is finding what you're going to talk about. So what I do is I go to, I go to Kaggle data sets, and I'll like, just search. I'll search like, environmental science or gender or whatever, and I'll see what data sets come up. Because I just want to, I just want to, I just want to see what's out there. And there's, like, this process of, you know, what is even available for me to look at, and then I go through, I don't know, 10 or 15 data sets just to see if there's anything interesting in them. And I'm always worried there's not going to be a light bulb moment, but there always has been, and there always seems to be something that like, oh, that's the right thing to do, and then we go from there. But the important thing is just picking something, so I have enough time to do it in the month. So I at this point, the day after the last show, I have no idea what's going to happen next month. I'll spend Monday, Tuesday night next week, looking through data sets for next month.
Rosemary Pennington
I wonder what you've learned about communicating science in different environments from working on this. So
Greg Matthews
I think in, like in the totality of so I did improv, I like took improv classes, I performed as improv. And this is stand up comedy. And sort of through all of, all of that, I think it's, I think it's changed the way I teach in in class, you know, from simple things that, like, where you stand when you're facing an audience, you know, to to, you know, not being boring, Right? Like it's if you just stand up there and talk about, you know, statistics. I know you and I don't think this is true, but some people think statistics is dry. And if you can take a dry subject and teach someone and teach someone that, but also keep their interest with you know, things that are at least, you know, funny to some people, it creates a good environment for learning. So like, the I'll give you an example where, when I would teach stat 203, it's Introduction to Statistics. Instead of, like, writing out examples, what I will do is, I'll be like, All right. Someone shout out, what do you want to do an example about and they will be like, All right, horse racing. And I'll and I'll sit there, and I'll make up an example, and we're going to do a hypothesis testing example, and I will make it about two horses or or two different groups of horses. They each run something, and this other horse runs something, but they eat a special kind of oat. And then you get to, like, make up a little story, and you're doing, like, a little bit of improv, and that gets them interested, because they get to choose what we're doing. The example about, it's all the same hypothesis test behind the scenes, but, like, that kind of stuff really works, right? 19 year olds aren't all that interested in hypothesis testing in general. But like, if you can get them to, you know, be involved in any way. It's, it's really helpful educationally. I think,
John Bailer
yeah, that kind of audience involvement sounds like an awesome strategy. I I often thought, when I was teaching, teaching stat classes, that that I was glad that the expectations were low coming in for many, you know, because it's, it's, you know, if you're, if you're teaching certain classes, the expectations are coming in sky high. You know, if you're teaching a geography of wines, there's a very different expectation than if you're teaching hypothesis testing and inference and but that's a that's an opportunity. It's not, it's, it's, it can be a blessing. So I, I really like that, and I, I'm curious, you've mentioned this kind. Of how the comedy and sort of that the improvisational has impacted your your thinking about this as a presenter and as a teacher. How about your consulting? Have you? Have you found that how has has kind of this, this work in comedy and in data art changed the way, or changed some of the ways you interact with with clients in a consulting setting.
Greg Matthews
So I don't actually, I mean, I don't know if it actually impacts that, but I will say that I've learned things through making data art and from doing analysis for the comedy show that I've actually used in consulting projects or in my research, right? I haven't used this yet, but like the most recent example of this is last night. I did, I did some change point analysis. Well, I didn't do it last night, but in the show last night, I presented some change point analysis using the pelt algorithm. I had never used this before, and I would never have come across it other than just I had the I wanted to use it for a comedy show, and so I studied the pelt algorithm for detecting change points because of a comedy show. That's incredible, right? I feel much more comfortable working with image data, right? So, like, when you do some when you do, like convolutional neural networks, or you're doing image classification, the reason I know how to do any of that stuff is because of the data art and working with images like as a hobby. And so I've actually learned quite a bit that helps me professionally, from doing these hobbies, right, from doing the comedy shows, or from doing from making this data art, because it's all, it's all coding, right? And so, you know that also brings me joy. I can justify doing it because it's professional work, right?
Rosemary Pennington
I wonder, as you've been preparing these guest lectures, was there ever a topic that you really struggled to just sort of get written in a way that you felt was funny and compelling and sort of what helped you get through that
Greg Matthews
last June. So June is pride month. Last June, we did LGBTQ health. And I am a straight white man, and I had to be very careful the way I wrote that. But I did. I think I did a very good job. I looked at, you know, I looked at, I did a statistical analysis of legislation in like Oklahoma, or legislation across the states. But Oklahoma has a lot of these that is trying to, like, you know, legislate LGBTQ topics. And so I did the, I, I just presented other people's work, or other people's proposed legislation on this, but that was a difficult topic because of the sort of sensitivity around it. But it was, it was a fun challenge, and I had to do it in a way that was aware of who I am and the topic that I was talking about. But I would say that was the most challenging topic because of what it was.
John Bailer
So what kind of recommendations might you have for people who are interested and in getting involved in data, art or or getting involved in and comedy?
Greg Matthews
So with a lot of things, like, if you want to be a lawyer, you got to go to law school, and you got to pass the bar, and then someone calls you a lawyer. If you want to be a professional athlete, you got to, you know, make a team and sign a contract. The bar to becoming an artist is you just saying, I make art now, right? There's no, there's no there's no entry, there's no gatekeepers. You can just make art, right? And it's the same thing with comedy. Comedy is a little harder, because you got to get people to show up. People to show up, to your to your shows, but like you can make art right now, right? There's no There's literally no bar to it. If you want to be an artist, all you have to do is say, I'm an artist now I make art. Just go do it. And I think people are afraid to fail, and they need to stop this. You're gonna fail a lot. Like every time you see someone who did something really successful, all you're seeing is their successes. Right? Whenever you see a comedy show, whenever you see someone do an hour long comedy special that took them months to write and they failed the whole time before that, what you're seeing is the final, finished product. Same thing with an artist. They screwed up that that piece 1000 times before they got the final thing. So don't be afraid to fail and just go do it. It doesn't have to be art or comedy. This applies to like everything, whatever it is, whatever it is that you want to go try. Just go do it, right? Just try stuff. It's okay.
John Bailer
That's good advice. That is good advice. And you know, in this we like to end with good advice, so I'm afraid that's all the time we have for this episode of stats and stories. Greg, thank you so much for joining us today. Thank you. It was an absolute pleasure. Yeah, thank you so much. Stats and stories is a partnership between Miami University. Whoops, I'm going to do that again. Stats and stories is a partnership. Between the American Statistical Association and Miami University departments of statistics and media, journalism and film. You can listen to us on Spotify Apple podcasts or other places where you find podcasts. If you'd like to share your thoughts on our program,