Check out a free preview of the full Building Custom Data Visualizations course:
The "Design Solution, Part 1" Lesson is part of the full, Building Custom Data Visualizations course featured in this preview video. Here's what you'd learn in this lesson:

Shirley reviews sketches from students and discusses their messages they want to convey in their designs.

Get Unlimited Access Now

Transcript from the "Design Solution, Part 1" Lesson

>> Speaker 1: So this is some of the sketches that we got, and I wanted to talk about. So the exercise was to look at the data exploration we did and figure out if there's any main kind of takeaways that we wanted to design around. And so think we got two of these.

[00:00:22] And so this one looks like Sankey diagram. Okay, so you're basically saying you are filtering down the actors and directors by a certain, you're not showing the whole dataset of actors and directors. You're filtering them down by maybe just the ones that have interesting correlations between the actor, director, genre.

[00:00:43] But that's a design decision you made and that's something you want to say in the footnote. So, I think this is super interesting. This is modeled after a sankey diagram. So sankey diagrams look like this. Well the one in D3, I think this is, the one in D3 looks a little bit like this.

[00:01:07] I haven't personally used that many sankey diagrams, but sankey diagrams, to my understanding, is really great for flows. So I've seen sankey diagrams used for looking at how people navigate between pages and the ways that they go or something like that. So it's really interesting to see sankey diagram applied to the actor, director, genre.

[00:01:30] I don't know if that's as much of a flow between each of them. And that's one of the things that's really great and unique about the web and this tool that we have. Which is that the interactions are possible and for the clarity, for making sure there's clarity in the visualization.

[00:01:48] I like that idea of being able to hover one of the lines and see as it goes all the way through from actor to director to genre. I think another thing that might be really interesting is at the very beginning of the class you said you were interesting in seeing these actor-directors and how they might contribute to revenue.

[00:02:10] And so If this can be visually fit in, maybe it would be interesting to say that each of these lines actually represent, the height of these lines actually represent the box office revenue of that particular movie that had that actor with that director with that genre. I think that would be really, really interesting.

[00:02:31] So and then we might be able to see if these sets of actors always work with this director in this particular genre. And I will also encourage you to sketch out a few other ideas around this central theme. And see which one works the best. But this would definitely be interesting.

[00:02:51] I think it's a really interesting application of a flow diagram. I don't know off the top of my head if it's the most straight forward, I guess, but I think it's definitely worth trying out. So that's really cool, thank you so much. And then we have this one, if you, Madeline, do you wanna talk about it a bit?

>> Speaker 2: Sure.
>> Speaker 1: And tell me where to concentrate on the screen.
>> Speaker 2: Yeah, we had a couple of ideas here. I think we were most interested in the top left and bottom right.
>> Speaker 1: This one.
>> Speaker 2: Yep. And that's just looking at IMDB score over the total IMDB vote count.

[00:03:35] And then the color association would be between nominations that movie got versus how many awards it actually won, which would be conversions, that would be wins.
>> Speaker 1: That's really interesting. I like that idea of looking at the conversions, I know we talked about IMDB score and binging by that.

[00:04:01] And then I see, I think, and then you added an extra layer of information. So you kinda overlaid multiple quote, unquote channels on top, multiple data dimensions on top of it, and now you can see convergence nominations. I like what you did down here where I see you sketched a little bit of what if it's not stacked on top of each other but rather side by side.

[00:04:26] I like that thought direction you went in. I do wonder though. So conversions, so nominations that go into winning, I feel maybe it's just because we just talked about the sankey diagram, but that could be a flow also. Convergence from nominations to winning and I wonder if that's an alternative that you can look at.

[00:04:49] And if there's other ways of representing this data, that would be really interesting too. And then the other one, I like, and you did do it. That is the sankey diagram you drew out at the bottom. You were trying to-
>> Speaker 2: We were trying to explore something besides a bar chart, I think.

>> Speaker 1: That's great. So what's the Co 1, Co 2 and stuff.
>> Speaker 3: There I thought break it out by production company and maybe see by company how well they're doing at converting nominations to wins.
>> Speaker 1: That's cool. Yeah. So this might be a really bad idea or it might be a really interesting idea.

[00:05:34] But one of the things I was debating about talking about, but didn't end up talking about is the concept of visual metaphors when you are designing. And so, my friend RJ Andrews is really, really great at this, infowetrust.
>> Speaker 1: Where he tries to always relate the,
>> Speaker 1: The visualization back to the data itself.

[00:06:07] So I think this one is the one I wanted to show. Profiling the parks, but it's a video so maybe we can watch it offline. But he is really amazing at tying the visuals back into the theme of the data. And so some of the simplest things that he's talked to me about, and given me feedback on my visualizations is something as simple as positive things that grow up, and negative things that grow down.

[00:06:37] Or good things that rise up, and that's something really simple, that's a visual metaphor. Things relating movies back to maybe the shape of your overall visualization end up kind of mimicking the movie data itself. Maybe the shapes are, looks like film records or like, but as long as it's not cheesy in a well done way.

[00:07:10] I think actually there's one that Nadieh did that I liked because it was so subtle. And she looked at the, I think top music that was played during, she lives in Amsterdam and they have a music that I think in December, they have radio that kind of just replays old classic, I think.

[00:07:38] And then so this is a visualization she did of last year's song selection, and when they were from. And one of the things that she did was that for the songs, I believe it was the songs that were played the most looked like a record. And then smaller and smaller and smaller.

[00:08:05] But it's something as simple as that, it looks a little bit like it. So you can see the record in here. And it's very subtle. But it's like a nice nod to what the data set is about itself. And so what I wanted to suggest, which could be a good idea, which could be a bad idea, is that considering the fact that we only have these two categories of won or nominated.

[00:08:31] And maybe it's that, and I'm guessing you mean the nominated one you only mean that it was nominated but did not win. And so slope graphs, I find them to be very good and concise.
>> Speaker 1: No slope charts, yeah. Slope charts are, I like them a lot because it's basically kind of relationship.

[00:09:06] You can see basically the relationship between two things and then tell from the angle of it whether it's going, if the relationship is up or down. And so maybe there's something that can be done where instead of a sankey, one column is the companies, and the other one is just very simple, if it goes up because it won, it goes down because it lost.

[00:09:32] And maybe there's something you can see there because of if there's a lot of them that win or if there's a lot of them that lose or something. Not suggesting it as a good idea. Just as an idea [LAUGH]. Because, again, at the end of the day, you have to see the data on the screen with your visualization to be able to say, this is good or this is not good.

[00:09:56] I think that's one of the really interesting things about designing for visualization. I feel like a lot of times when we design for other things, if we design for web, like a web page, oftentimes we can just design in Photoshop or in Illustrator or something. And we know exactly how that page will look like, but when we design for visualizations it's very, very rare that our initial design that we sketch out turn out to be a good one.

[00:10:25] Because it's so dependent on the data and the shape of the data. And even if we had done all of the data exploration, maybe sometimes we miss it by a little bit. So the design process is a lot of iteration over and over again, of seeing the data on the screen in the design and modifying that, or going with the different design.

[00:10:50] And this one, I like this one.
>> Speaker 2: So this one is a similar idea where we have nominations over, sorry, wins over nominations, which is the shaded area. But it would be seasonally, so you could see how many movies were nominated in January for example versus December.
>> Speaker 1: And then you have this that, this is super interesting and I like it very much which is that it's angle is month but then as you go out in radius it's year.

[00:11:26] And so my concern with this would be if you have so many years, it might be harder to see. It might be harder to be able to compare the win versus loss, because I'm guessing it means the orange ones are the wins and when it's not colored, it's losses.

[00:11:48] But if you have so many years stacked on top of each other, it might start getting overwhelming to be able to see those trends. But what this does really remind me of is, there was a visualization by Truth and beauty which is more Stefaner and some of his associates in Google and use lab.

[00:12:16] And it looked at the search interests for different keywords for across the years. So it establishes in 2004, yellow is 2004. And then blue is 2016. And the x axis, so they actually encode, they do this where they encode the data attribute of date to both the x axis and the color.

[00:12:45] So, because they want to make sure, make doubly sure that l you pay attention to the date. And then the y axis is the search interest. So how many people were searching for it at that given time. And what I want to show you is, I believe. So this is one of the results.

>> Speaker 1: And they do this.
>> Speaker 2: That's cool.
>> Speaker 1: So this is one of those examples where they use animation. To show, actually I think I'm, actally this isn't the right, yeah I think it's this one maybe. So play, and this is how they use animation to show you the passing of time.

[00:13:42] And so you kind of get. So, for example, this is aparently, this search interest for peach on Google. And you can see, again, they kind of hint to you that both color and exposition denotes the So yellow, like in the beginning there wasn't as much, but then like It really builds and builds outwards like that where there seems to be increased search interest in recent years.

[00:14:11] There's other ones that they have that's super interesting where I think it's words that are trends.
>> Speaker 1: That really, really spiked up in a certain year and then came back down I think so-
>> Speaker 2: You think acai that was like a really trendy fruit in like,
>> Speaker 1: I think they let you.

>> Speaker 4: Do a search for that and then I'm come back.
>> Speaker 1: Find a food, yeah, acai. No, they don't have it.
>> Speaker 4: Maybe it's not a fruit. No one knows what it is.
>> Speaker 2: Try again from Brazil.
>> Speaker 1: Charts that look like the food okay, that's awesome. But let me kind of there was one, wait, wait.

[00:15:11] I remember what's that Starbucks drink? Pumpkin spice latte, I think they have one that was like, they have, so in 2006 and 2007 in all of these it's not that big of a deal, and then 2011, and then it just keeps going up and up and up. And then there was an There was another one that peaked at 2013 or 2012 or something, and it's been coming back down since.

[00:15:50] And so I really love this one as an example of showing how they encode the data, not to just, they kind of, that radial format reminded me of this. But they also kind of take advantage of the fact that seasons, months are kind of circular and cyclic and they made this animation out of it.

>> Speaker 1: So maybe this is a direction that that can go in. So maybe instead of, maybe it's that, cuz this is the search interest of 0 to 100, but maybe your radius could be the percentage 1 versus lost. And that's how you graph that. And that could be a really interesting option.