Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #32: Teaching Data Visualization
143 Plays9 years ago
Welcome back to the show. In this week’s episode, we’re talking about teaching data visualization. What are best practices? Are there great exercises and assignments to give? What should students walk away with at the end of the day, week,...
The post Episode #32: Teaching Data Visualization appeared first on PolicyViz.
Recommended
Transcript
Introduction to the program and experts
00:00:00
Speaker
This episode of the Policy Viz podcast is brought to you by the Maryland Institute College of Art. MICA's online graduate program and information visualization trains designers and analysts to translate data into compelling visual narratives. Join expert faculty such as Andy Kirk, Marissa Peacock, and John Schwabisch to mine the data and design the story. For more information, go to mica.edu backslash MPS in Viz.
00:00:38
Speaker
Welcome back to the PolicyViz podcast. I'm your host, John Schwabisch. I'm excited for today's episode because I'm going to try another group episode where I have count them four guests to help me talk about strategies and ways to effectively teach data visualization. Instead of introducing them one by one by myself, I thought I'd just let them have the honor. So let's start with Robert Cusara from Tableau Software. Robert, floor is yours.
Teaching data visualization: Insights from industry and academia
00:01:03
Speaker
Hi, this is Robert Cusara from Tableau Software.
00:01:06
Speaker
I'm a research scientist and I used to teach at a university. I no longer, I'm not actually doing this, but I still have opinions on that. And I used to teach in a computer science department and used to teach undergrads and grad students and of course PhD students. And now I'm on the research side in industry, so I'm still interested in what people are doing, but I'm kind of no longer actively involved.
00:01:31
Speaker
Hi, I'm Marty Hurst. I'm a professor at the University of California Berkeley in the School of Information and also affiliated with the Computer Science Division. And I'm an academic. I primarily teach master's students in our professional program. They are learning information science and technology. They're really interested in being practitioners in the field, although I also guide
00:01:57
Speaker
research students, PhD students, and occasionally teach undergraduates as well. Hi, I'm Tamara Munzner, and I am a professor at the University of British Columbia in the computer science department. Historically, I have taught a grad class within computer science, but that's open to people from other departments. Then recently, I've tried some experiments on teaching vis to a group of journalists who are not programmers.
00:02:22
Speaker
And next fall, I will start in on teaching professional masters to data science students who have some programming and are much more oriented towards practitioner than research track as my class has historically been. And hi, I'm Eitan Adara. I'm an associate professor at the University of Michigan in the School of Information, which is my primary appointment, but I'm also in the computer science program as well.
00:02:47
Speaker
So I teach a class on information visualization. I guess mine is a little bit odd in that it's the only info of his class at the university at the moment. And so it's cross listed both within the School of Information and Computer Science. It's a graduate course, mostly master's students who are trying to become practitioners.
00:03:06
Speaker
in HCI broadly, but there's other things as well like people who are going into health informatics or education. I occasionally get people from urban planning and other schools as well. They simply, again, because it's the only game in town as far as a class. There are sometimes PhD students, but again, it's mostly practitioner focused.
Core skills and programming languages for data visualization
00:03:28
Speaker
And so for those of you who don't know, I'm not just a podcast host. I also teach data visualization. I teach a smattering of classes at the Georgetown University Business School, which tends to be sort of a two-week class. I also teach in the Policy School. I also teach an Exec Ed class there. And I also teach at the Maryland Institute College of Art, which is an online data visualization class, which has its own sort of separate challenges, I think.
00:03:52
Speaker
And then I also teach public and private workshops open to the public and then also for clients both full day and half day and full week classes and the whole lit. So now we have a good smattering of folks doing a lot of different great stuff. I want to start by asking, we're audio here, so we'll see how this is going to work. I want to ask you, what do you think are the most important skills to teach in a data visualization class? And again, sort of.
00:04:17
Speaker
Acknowledging that we're each teaching different groups of people with different skills coming in and different demands that they'll have when they go out into the world. Let's start with Marty. Marty, are there certain skills that you think that people should have when they're leaving your class at the end of the semester?
00:04:32
Speaker
Well, it's always hard to say most and so on. So I actually talk about how we're going to learn. I'm very into pedagogy, which is the science and practice of teaching and learning. So I talk about how will we learn information visualization at the start of the class. And I'm using four P's now, principles, practice,
00:04:56
Speaker
peer learning and review and programming techniques. So we do emphasize programming in this particular class. And practice, I mean, getting lots and lots of practice. Because as we all know in design classes, if you don't practice, you don't really learn how to do it very well. You have to look at a lot of examples and actually produce a lot of examples yourself. We also want to learn the foundational principles.
00:05:21
Speaker
And there are a lot of great foundational principles for visualization, both cognitive and perceptual. I'm afraid I'm going to steal everyone's thunder, sorry about that, for those who follow. Something that I'm doing that I'm very excited about is peer learning, where students talk to each other even during the lecture. I pause frequently and let them discuss the concepts and do short design exercises or short analysis exercises during the lecture, like a couple of times during every lecture period.
00:05:50
Speaker
And also they do exercises before each class begins individually to prepare for class. And then we actually learn four different software tools during the course of the semester that correspond to the four main units that we study, which are the basics, the perceptual basics, which are reinforced with producing a static design with a kind of a basic JavaScript tool. Then we studied narrative.
00:06:18
Speaker
how do we create a narrative expression of a story, which actually was pretty new a few years ago. And now all the students are, many of the students are asking for that. It's become quite a more standard thing to do. Then we do exploratory data analysis, which is, you know, it's opposed to just presenting static information. It's how do you use visualization to explore and learn about a dataset and answer questions. And then finally, you know, multi-dimensional visualization animation, those advanced topics.
00:06:46
Speaker
We learned a modern visualization tool for that, D3, to express those concepts. And then they put everything together in a final project, which they end up combining these ideas in really interesting ways as a capstone.
00:07:00
Speaker
So that's a long answer to your question, sorry. No, that's fine. So you touched on a few things and I think what I'd like to do is take some of those sort of piece by piece and farm them out to each of you. So I want to start by talking about the programming languages and what programming languages, the ones that you are teaching and the ones that you think are obviously most valuable. So let's start with tomorrow. Tomorrow, are you teaching JavaScript and D3 and you have CS students. So that's sort of a different skill set already coming in.
00:07:24
Speaker
Yeah, it's actually a sufficiently different skill set that I take, ironically enough, a very different approach, which is I don't explicitly teach any of these tools in my standard grad class. So my own class is very, very heavily focused on principles and analysis. And then there's this major component, which is a final project. And in that they can use a language of their choice. And I don't mandate it. I also don't teach it.
00:07:49
Speaker
So they're expected to pick that up on their own, and that's one really big difference from the way I run it than the way Marty has described. Now, when I have recently started moving outside the CS department, then with the journalist class, I tried doing the entire thing in Tableau. With the data science class, I think we're going to be pretty focused on R, and then a lot of the tools that Hadley Wickham has built, like ggplot and ggfizz.
00:08:15
Speaker
So what's happened is a lot of my students used to do a wide variety of things. In the last few years, they've all voted with their feet, and it's like 90% D3. So D3 is incredibly popular. Can I just intervene for a moment? This is Marty. I forgot to say we use Tableau for the exploratory data analysis. And we use D3 not because I think it's maybe the best pedagogically, but because the students all demand it. So the same thing that Tamara said. It's really the demand in that case.
00:08:42
Speaker
because it is a difficult language to learn. Right, and so when you're teaching D3, are you also walking through all the HTML and CSS background that people need? We start with that, yes. And Eitan, what about you? What tools are you teaching in the information school?
00:08:56
Speaker
So they're welcome to use whatever they want for their final projects and they probably mostly end up doing D3. I insist on having something implemented at the end that's interactive and it's one of the few games in town that's actually handy for that in particular because they want online portfolios and so
00:09:14
Speaker
they pick something where it's easy for them to generate at the end and so what i've had to do and this is over many years i used to teach like one brief lecture that introduced them to d3 or whatever other tool and that wasn't enough and now it's become sort of five one-hour lab sessions um at the beginning of the semester so four of them are d3 various elements of d3 and one of them is um tableau
00:09:39
Speaker
This isn't ideal, but again, because of the high variance in the students and their backgrounds, some of them are coming from the CS program, some of them have CS background, but some of them are coming from English Lit, where they've had some programming as part of the earlier classes in our program. But because I insist on them actually building something, I have to give them enough resources basically to make that happen. We've settled on D3, I've tried other things, processing, P5,
00:10:08
Speaker
and D3 seems to be the one that's stuck. Lots of examples, which helps and hurts, but yeah.
Tools and methods for data exploration and visualization
00:10:17
Speaker
And so Robert, when you're talking from working at Tableau, when you're talking with people and instructing them on how to use Tableau, are you finding that the people who want to learn Tableau have, or maybe they don't have the skill sets at that time to do D3 or to do R and they're just more interested in the off the shelf, drop and drag, be able to create visualizations in that way?
00:10:38
Speaker
Yeah, yes and no. I mean, it's a very different approach. In Tableau, you can just play with your data. In D3, you do a lot more infrastructure, you have to do a lot more hanging certain JavaScript things from certain calls and actually making those things happen. And it's actually fairly difficult. So the hard part in my experience teaching D3, and I used to teach, I actually started out teaching people to do Java, and then
00:11:08
Speaker
switched to JavaScript and Protoviz many years ago, and then D3. And what I actually found was a lot more important to teach them than D3 and HTML and CSS stuff was JavaScript and how to use closures in JavaScript and why certain things work certain ways. Because D3 is very clever in the way it uses JavaScript and JavaScript functionality, but it's also hard, especially for people who aren't really all that used to functional programming and to more advanced programming.
00:11:38
Speaker
And of course, there's a lot of stuff out there. I mean, Microsoft has been amazing in just creating this whole ecosystem with lots of examples and lots of forums and places we can find help now. But it's still a lot of programming involved. So with Tableau, it's just easier to get started.
00:11:54
Speaker
They really focus on the data and I know a lot of people who actually do their data exploration in Tableau and also the data aggregation and then when they know what they want and they see the things they want to actually build, they then export the data from Tableau and that also already does the data modeling and stuff for them.
00:12:11
Speaker
and then get that into JavaScript and then make it load the CSV file, whatever that comes out of Tableau. So you can combine those tools in many different ways. And of course R as well, I mean it's the same thing. R is a great tool for, especially for people who are really fluent in it, but getting to that level takes a while.
00:12:29
Speaker
I just wanted to add this, Marty, that I guess Hadley Wickham is making a wrapper for R to make it much more user-friendly, so that's probably going to raise the profile of R, especially in the whole data science movement. I think it's going to ascend in some ways soon. What I find interesting hearing all of you talk about the tools is a lot of focus on interactivity, and R may be the exception to that.
00:12:53
Speaker
Are you focused on making interactive visualizations because that's what tends to get shared and gets the highest profile? Or because that's just the way you think people should be learning or that static or your students don't need to create static graphs. I didn't hear anybody really mentioned Excel or even sort of creating more stylized things in Illustrator. We do teach that stuff. And this is a, you know, 14 weeks for me. It's a semester long class. And so.
00:13:16
Speaker
We go through the static stuff initially, like those are the first few lectures and we don't necessarily show them tools to do that. They're familiar with Excel mostly, my students anyway, and Tableau gives them some of the resources so they're able to generate.
00:13:30
Speaker
the static stuff early on, and we go through various design philosophies around that. So we don't ignore it, but at least for my purposes, interactivity is a key focus of information visualization production, basically, the artifacts that we produce at the end. And so we do spend quite a bit of time, and D3 offers those abilities at the end. Although there's other options now, but D3 is one of the few.
00:13:59
Speaker
This is Marty. I actually spend a really good proportion of time on the static stuff. So the high charts is a JavaScript package that's pretty much static. I mean, there's some dynamic parts, but it makes very aesthetically pleasing static visualizations with a little bit of dynamics. And then we do Illustrator. We introduce the students to Illustrator for the narrative infographics, actually. And then Tableau is the introduction for the dynamic for the exploratory data analysis. But I think it's like an artist, they practice sketching.
00:14:29
Speaker
extensively before they get to the painting. And I think you really need to understand the static aspects of visualization before you do the animations and the dynamic parts. Yeah, that's really interesting. And are all of you teaching other technology and platforms like GitHub and some of the other sharing tools out there? Not in this course. There's other courses for that. Okay. Yeah, same for me. Ditto. Interesting.
00:14:54
Speaker
So before I want to move on to practice, but before I do so, I want to turn it tomorrow for a moment because tomorrow you mentioned a little that your class is a little more focused on perhaps the more of the academic research side and a little bit more of the theory. And a lot of what we've talked about so far is using tools to sort of do practical creation. So I was wondering if you could talk a little bit about how your primary course sort of differs from what the rest of us might be teaching.
00:15:17
Speaker
Well, one thing is I'm intrigued actually both by this discussion and by the earlier panel at VIVS to think about changing how I teach. But the way I currently teach, it's quite focused on reading of principles. So there's a book that I recently finished which we use as the core
00:15:34
Speaker
In the first two thirds of the class, for every lecture, they read one chapter from the book and then one research paper because one of the goals of the course is actually to learn how to read research papers, which is a skill that actually needs explicit teaching because they're kind of a different beast than what people are used to doing.
00:15:52
Speaker
One of my goals is really to kind of, even if they don't continue in visualization, I want to develop them as researchers. So they do reading of research papers and the books, then they also do writing of comments about both the paper and the chapter, which they turn in before class. And then what we do also more towards the end of the class is they read a research paper on their own and present it to the other people in the class, which is both about building their skills for reading and then also for presenting.
00:16:23
Speaker
And then a lot of the practice component comes from this big, fairly monolithic final project where we do a lot of checking in along the way. So I'm interested to move to a world that might have more of these in-class exercises and explicit teaching, although it's kind of hard to imagine. I can imagine if I started doing an undergrad class, I might be able to do a lot more of the practice things built in. But it's a somewhat different goal at the grad level than the undergrad level.
00:16:49
Speaker
at least given how we structure our program, one thing that's unusual about Canada is the master's level is typically actually a research master's where they do research rather than just being a course driven master's as is common in the US. So in some ways our master's students sometimes act more like PhDs in the US which is a different flavor of things as well.
00:17:09
Speaker
So you've all mentioned final projects as part of the assessment, and you've all sort of mentioned practice and exercises, and I'm curious if you have specific exercises or assignments either in class or as homework that you really think get the students really engaged and get them to learn. So maybe I'll start with Marty since he sort of laid out those four points at the very beginning.
00:17:30
Speaker
Well, I could go on all day, so I have to, I have to choose. I actually just got a paper accepted at the CHI conference on variation on an exercise. I don't want to get too researchy for your audience, but a lot of us are interested in how do we bring teaching online and where you have a large number of people taking the course where the instructor can't necessarily give feedback on every assignment. And so a lot of us are really intrigued by that question and I am as well.
00:18:00
Speaker
And...
00:18:02
Speaker
So something I did in my class the last two years was I had students design a visualization. I gave them a data set and I said, design a static visualization using high charts to be able to answer a whole host of different questions someone might ask about this data. We studied the principles for that and so on. And then I said, we're going to put this on a crowdsourcing platform and see how many questions people can answer correctly about your design.
00:18:32
Speaker
And so that was very exciting because the students could see if people were able to answer the questions correctly. And we kind of got a spread on the results. So this summer I actually wanted to compare what happens if people answer questions versus if people if actually I want to do this with students have them answer questions about each other's designs.
00:18:52
Speaker
and compare it to applying guidelines, which is another standard technique used in human-computer interaction. It's less commonly used in infoviz, though. It's called heuristic evaluation, this technique. And so I did that, and I had students do heuristic evaluations on some designs, some infoviz, and I had students answer questions about those same visualizations. And I wanted to see kind of which was more accurate at determining
00:19:18
Speaker
which designs were best. But I didn't really know which designs were best. It's sort of a chicken and egg problem. I thought I knew which were best, but I don't really know. None of us really know which designs are best in some sense. And what I found was we got really quite a good correlation between these two ways of measuring designs with a few outliers.
00:19:39
Speaker
And that made me realize that having these two radically different ways to assess our visualizations might be very helpful for students because the professor might just say, I think this design violates principles. This design is good. And the students might be saying, well, who are you to say? Yeah, yeah.
00:19:58
Speaker
They don't always like that, right? And sometimes as a professor, I don't really know for sure if the student's design is good or not. And so if people answer questions and they get them right or wrong, that's another way to assess the design, but maybe I didn't have them answer the right questions.
00:20:13
Speaker
Maybe my question set was wrong. So having these two ways to assess simultaneously gives another check. It's sort of checks and balances on the design. And I found, like I said, a good correlation between these two. But then when there was an outlier, that suggested that there was something special about that design. Maybe people were able to answer questions really well, but I or the students
00:20:37
Speaker
said it wasn't a good design, it deserves a second look. This is what happens when there's automated grading, like automated essay grading. People worry that there's a brilliant essay that the computer says is bad. I'm very excited about this sort of approach moving forward of thinking about how do we have peers assess each other's designs in more innovative ways
00:20:59
Speaker
so that we can do it at scale, but not have people, you know, avoid the danger of some brilliant design falling through the cracks and being erroneously marked as being a bad design. Right. Yeah, interesting. Eitan, what about you?
Innovative teaching methods in data visualization
00:21:13
Speaker
Do you have favorite exercises and assignments? I do. I guess I should give a little bit of context, though, so it'll make more sense. The way I structured my class,
00:21:23
Speaker
is there are two major assignments. So one is an individual assignment where they build a visual explanation of the sort that Brett Victor does, where you're trying to explain some scientific or mathematical concept and constructing a visualization to help that. The second one is a large group project at the end of the semester.
00:21:42
Speaker
But all throughout the semester they're actually practicing the construction of visualizations. This is my favorite part and I think the students react to it as well. But basically what I do is I flip my classroom so they watch a bunch of videos in advance of class.
00:21:58
Speaker
Probably about a half hour's worth of video lectures. They come to class and we go through this process where they're split into groups at random. I talk for like three, four slides about some new visualization that they haven't seen before. I ask them a group question, they think about it, they answer the question, we talk about it, and this repeats for a little while.
00:22:19
Speaker
And then at the end of class, I give them a project related to the topic of the lecture. So let's say it's about hierarchical visualization. The topic might be visualize a tennis match for me. And so I already have a visualization in mind, like I found a research paper or some idea that I like that I found outside, but they haven't seen it. Like this is not something they've looked at, nothing they've read. It only vaguely has to do with the topic of the day. It will draw on the things that they've learned.
00:22:49
Speaker
But then they basically, in these groups, will go through the exercise of designing a solution. But they haven't seen the actual, you know, professional, in quotes, solution to this. They have to work through it for about an hour and a half, iterating over a number of steps, like it's a structured design exercise.
00:23:06
Speaker
at the end, each group produces their own view of what that visualization should look like. And for the next lecture, they're tasked with reading the paper, you know, the professional solution and comparing theirs
00:23:20
Speaker
to whatever the professional had built. And I found this to be highly useful because they have to think through all the design challenges, like the wicked design problems that we talk about in visualization. Whereas if I didn't have them do this exercise, all they would do is sit there and complain about the mistakes that the professional had done. Like they hadn't thought through the challenges, the trade-offs, like everything that you would have to do to make this work.
00:23:46
Speaker
And so I like this exercise. It's been quite effective in forcing them to think through the different aspects of the problem. So what is the domain task that the people have? What is the abstract task? Like breaking it apart in that way. And I think that's been a great thing. Like this is part of my favorite part. And I think the students really resonate with us. But again, to make this happen, I've had to shift stuff around, like doing flipped lectures so I could move some of the materials out of class to make this work. Right.
00:24:14
Speaker
Robert, I want to throw the same question to you, but I also want to ask a secondary question, which is, as someone who's coming from Tableau, does Tableau view it as part of their role to provide datasets for people to use as they're trying to learn the tool for instructors to use to help other people learn the tool? Well, I wish we did that a bit more than we do, actually, because we have some of that. So we have some datasets. There's a page somewhere in our forums for educators, and it has some links.
00:24:43
Speaker
But we definitely want to do a bit more of that as well. So there are data sets. There are some materials. But I'm actually pushing to get more of that stuff on there to make it easier to share these things. And certainly, I know that people do really interesting things in their classes. We haven't really collected those or tried to kind of come up with a list and point to them.
00:25:05
Speaker
But I think what I found, I mean, I was doing the things when I was teaching that I guess everybody else is doing, have people have the students find data sets, look for interesting things to talk about, do a big presentation at the end, perhaps, or something like that. What I found was really important, though, was to make sure that they stay on track and keep the eye on the price and don't wander off and just start making pictures.
00:25:30
Speaker
because I found that a lot of the students ended up just playing, which is good, they should play. But they need to play in a way that actually gets them somewhere to an interesting insight, not just to, hey, look at this colorful thing that I just made. And that actually came up quite a bit more often than I thought it would. So it's interesting to me that that's actually something that you need to keep reminding the students of, at least I needed to do that when I was teaching.
00:25:58
Speaker
And so that's a really good question. I want to broaden that to everybody. When you're asking students to create their final project or create a homework assignment or what have you during the course of the semester, are you laying out for them what their target audience should be? Are you asking them to define who the audience was that they had in mind? Or are you just letting them sort of have the world as their oyster?
00:26:15
Speaker
Oh, this is Marty. For final projects, they have to define an audience, for sure. And actually for the narrative infographic as well. And in fact, even in practice, I have an assignment due on Wednesday where they have to critique a design. And one of the questions is, who is the target audience of this design? So we definitely emphasize that.
00:26:36
Speaker
Right. Yeah, same here. There's a worksheet that I make them go through in their proposal and their sketchbook, which they have to keep modifying as they go. And the question at the top is, who's your audience and what do they want to do with your visualization? Right.
00:26:51
Speaker
Tamara here. One thing I found in my class is, as I mentioned before, we have the big, big project at the end. And over the years, a lot of the students have gravitated towards the kind of project that we call design studies in visualization, where there's this idea that there is a specific group of people that you're designing the visualization for. And a huge amount of the challenge in the project is to translate from the domain-specific task and data of those people to the abstractions that you can actually
Interdisciplinary approaches to teaching data visualization
00:27:19
Speaker
use in visualization to try to make sense of it.
00:27:21
Speaker
And so a big big focus of these projects has been exactly this process of abstraction, which is a way of understanding for what are the audience's needs and how could we address them in a visualization system in some way that might actually be effective and what I found that.
00:27:39
Speaker
is in earlier years there was more of a mix between technique-oriented projects and design study projects, and I can't tell whether it's that the way I've taught has changed or whether the appetites of the students has changed, but these days it's very, very heavy on the design study style projects. Interesting.
00:27:57
Speaker
So I want to ask a final question that's a little bit broader. I also, I don't think I have time, I want to talk about sort of best books and blogs and resources, but maybe what I'll ask you to do is send me your favorite and I'll put them on the website. So instead of asking that specific question, I want to broaden the conversation here a little bit.
00:28:12
Speaker
that a lot of things we've talked about here, the data visualization classes that each of you are offering are being offered in different disciplines and different departments and sort of teaching lots of different skills for lots of different audiences. And I'm wondering if you think schools, and Aton for specific examples, the only class at the University of Michigan on information visualization. And so I'm wondering whether you think schools and universities should take a more interdisciplinary approach.
00:28:37
Speaker
to teaching infoviz or teaching dataviz and not having them sort of separated into these different departments. I'm teaching in the business school and I'm teaching in the policy school. Should there be a central place where people come together or is the way that schools are organizing these classes the way they should go about it that there's a different sort of demand in a policy school than in an information school? So A-Time, why don't I start with you since you are the sole provider of education there.
00:29:05
Speaker
I would say that it's a little bit challenging to sort of address the needs of all these different constituents, all these different student types.
00:29:12
Speaker
There's stuff in here that I think applies everywhere, like I teach the perceptual stuff and the psych stuff and the Gestalt theories and all that. And I think that applies no matter where you are learning this material. The examples that I might pick might be different depending on whether I was teaching a computer scientist or somebody in the business school or policy or whatever. But sort of the core is the same.
00:29:36
Speaker
When it gets to the projects and what they can build and how they do it, there there might be some reason to split it up. You have the CS students who are quite technically adept, who can build very quickly and that's what they want to do. And you have the more UX design students who care more about the interface and what it will look like and the experience and so on. Now, on the other hand,
00:30:00
Speaker
Like mixing those students up in the same class has actually been beneficial in some ways. Having a group that's mix of programmers and designers or domain experts has led to some interesting projects. So for the final projects, they're able to bring in data sets that I may not have thought of. So I've had people from urban planning come in with data sets or from public health come in with data sets.
00:30:23
Speaker
the students all sort of bring different things to the table, and I think that's interesting and exciting for them. So I'm not sure if I would argue to split them up, I guess, at the end of the day. Like, I think that there is some benefit in having that kind of interdisciplinarity. On the other hand, would I like some help in teaching this, of course? It's a lot of work to teach to all these students, but I really do appreciate the fact that they're all sitting in the same room and can interact and teach each other something.
00:30:52
Speaker
Marty or Tamara, Robert, any thoughts on that? This is Marty. Well, I used to teach the only one at Berkeley because I taught one of the first of his courses in the country in 98 and it was more researchy. I mean, I was sort of making it up as I went along and it was mainly just research papers. There weren't many tools to use. And then we hired somebody in the CS department who could teach a CS version.
00:31:16
Speaker
And so we alternated semesters, but we pretty much taught the same course, except the students did much more of an implementation project. And it was more researchy on his side. So that freed me up to teach a more of a practitioner course on my side. So basically, we made it a division of labor, sort of not just tacitly.
00:31:37
Speaker
And it made it easier for me because I wasn't trying to please two constituencies with different technical levels. So the content was pretty much the same, although I had fewer research papers. And as the years went by, I made mine more and more practitioner focused. And I think it made a much more satisfactory experience for both sets of students. That professor has since left Berkeley. So right now, there's nobody teaching the research version, which is not a good thing. I'm sure that's going to have to change.
00:32:04
Speaker
But the final point is there's this data science, whatever you want to call it, surge, all across campuses. And there needs to be a more global course taught at the undergraduate level that covers some version, a lighter weight coverage of it, which might actually be done in R, this new wrapper around R. I think actually people are doing it, information visualization in every social science now.
00:32:33
Speaker
And I think what determines whether or not there's a course taught in a department is whether or not they have hired a recent faculty member who knows how to teach it. And that's really going to be the term to determine it going forward. So I sort of have the mentality of let a thousand flowers bloom in the sense that I think if there are multiple people on a campus that want to teach it, then, you know, by all means, and they will bring different things to the table.
00:32:56
Speaker
And it's interesting hearing about how Marty's diverged from the CS department version in a very deliberate way once there were multiple people teaching. One of the things I do in my own course is I actually allow different kinds of projects, so I know a lot of people emphasize that in their final projects they mandate that the students program. I actually have programming projects as one option and then analysis projects as another option where they can use an existing tool to do a very detailed analysis of a data set
00:33:24
Speaker
and that's frequently used by people that are outside CS that are non-programmers or they might compare and contrast two tools and the kinds of insights you can get from them.
00:33:33
Speaker
And that's one way that I make things accessible to people that are non-programmers is just give them a choice of what to do with that. The other thing also in responding to this idea of data science getting really popular, I've been thinking pretty hard about how to refactor what I've normally done in a three-credit graduate context and figure out if there's a couple of one-credit modules that need to fit in in a lot of different contexts, how can I sort of
00:33:56
Speaker
grab out the core of that and teach that in separate ways. So that's something that I think a lot of us are going to be doing as the data science stuff hits. And I'm sort of well on my way of thinking about how to extract
00:34:10
Speaker
What I often think about is that there's core visualization ideas, and then there's what programming environment are they going to use? Is it going to be R or Tableau or D3? And then what domains might you be targeting? You know, journalism versus genomics. You might have some very different examples that are both trying to illustrate the same underlying principle. So this question of how can you have these sort of retargetable vectors of modules of visualization is something I'm thinking about a lot these days.
00:34:35
Speaker
Yeah, interesting. I really like that. On a, I guess somewhat related note is we have within our curriculum at the School of Information, we have some belief about cross cutting things that should be taught, statistical literacy, security, privacy, ethics, whatever those are.
00:34:53
Speaker
concepts that should be taught across all classes and one of the things that we've talked about but haven't implemented and I'd like to see more of is a presentation and communication sort of thread where the belief is that in most classes there's an opportunity to teach something about visualization and if it's possible to sort of have these modules as Mara is describing I think that's a great idea. You know that having those available to faculty to pick up and integrate into their classes
00:35:20
Speaker
that are teaching other things, I think that would be awesome. And doing that beyond just the small department or the school and having that available at the university level would be great. Because I do think that there are opportunities here, visualization being a great example, that sort of affect so many different areas and fields. And being able to teach at least parts of that in different classes would be great, from my perspective.
00:35:42
Speaker
And just to build on that in the spirit of broadening things, this is Marty, I do worry sometimes that in this data science enthusiasm that just general, the humans are getting lost a bit. And it's, Infoviz is just one part of the broader HCI human-computer interaction, you know, thinking presentation is a nice way that Aton just put it. And that it's really about how do we make things understandable to people and taking
00:36:09
Speaker
people into account and so I really do hope that it's HCI more generally with the infoviz as the data specific feature that gets incorporated as we think about this more broadly.
00:36:25
Speaker
These are great ideas. And I think fodder for more discussion. Unfortunately, I think we're going to have to stop because people are going to be arriving at work right now and they're listening to this podcast. So I want to thank each of you for coming on the show. It's been really interesting. A continuation of our panel in Chicago at the Viz conference. And I trust we'll be talking about this more as we go through this semester and further semester. So thanks to each of you for coming on the show.
00:36:49
Speaker
Thank you. Thank you for having us. And thanks so much for everybody for listening. Hopefully you've enjoyed today's episode. If you have comments, please hit me up on Twitter or on the website and please rate the show on iTunes so others can find out about it. And until next time, this has been the policy of his podcast. Thanks so much for listening.
00:37:18
Speaker
This episode of the PolicyViz podcast is brought to you by the Maryland Institute College of Art. MICA's online graduate program and information visualization trains designers and analysts to translate data into compelling visual narratives. Join expert faculty such as Andy Kirk, Marissa Peacock, and John Schwabisch to mine the data and design the story. For more information, go to mica.edu backslash M-P-S in viz.