[MUSIC] Welcome back, these next sets of lessons allow us to get into pre-attentive attributes for data visualizations. We're now going to talk about how to make visualizations more usable, pr rather more accessible for the reader. When you see a poorly designed visualization, who do you think she get to blame. Should we blame some nebulous source like the data? Which is often who we blame, it's like just the data. The data doesn't show it, but what if someone has trouble interpreting a graph that you created? Is it the reader or audience that is at fault? Is it basically that the reader just doesn't understand what you're trying to tell them? You're right about that, but the answer to who is to blame is the person who designed the visualization. If the reader can't understand it, that means that the person who designed it, did not do a good job of designing the visualization. So when you're doing visualizations think through your designs, and take into account the needs of the user of the visualization. You're the expert on the data, they're not, and that is the focus for this lesson. We'll examine a couple of use cases to define the point of reader usability. I want to use an iconic example to explain what I mean about this accessibility usability thing, and this iconic example is the London Underground map. Before 1933, the map was based on geography, it makes a lot of intuitive sense and some of the lines go quite far but some of the lines are much shorter. And there was a real problem, there are a lot of lines in Central London and there are other lines that were very far from Central London and so it actually couldn't even be shown on the map. I have it on the screen now in front of you, so lines in Central London were squeezed together, making it hard to read. But lines further out couldn't even be shown on the map. So even though it was in a geographic representation, it wasn't actually very good at showing geography either. Harry Beck who was an employee for the London Underground in 1933, thought about this problem a lot. Passengers wanted to know where to transfer to get to where they needed to go. So geographical accuracy ironically was not important at all. They just need to know how, where their station is versus the other stations in the London Underground. So he designed a map to be almost like an electrical schematic instead of a traditional map. And now even to this day, this is what Londoners use to navigate the subway system, even though there's many more lines. It is a map that has nothing to do with geography, except in reference to the particular stop versus other particular stops in the London Underground. It's a very innovative map, and it's iconic you see it even in other countries as an example of a way a system should be drawn. The key to why this redesign works is that it conveys the information in the way the users need, to eliminate clutter and distractions. It's not geographically accurate even though it's a map, but it turns out that it doesn't matter, what matters is that it's accessible, and usable. So we're going to channel Harry Beck in a very small way, to make some modifications to the visualization we were working on in the last lesson, so let's go to it now. So let's review the question that we had in the last lesson as a refresher. The question that was posed to us is, why sales to certain customers, on certain products, aren't profitable despite there being a lot of sales. From our last lesson, we as the experts of the data looked at discounts as a possible reason. So now we're going to apply our expertise about the data towards making our visualization accessible. And thus answered a question about whether discounts are responsible for this. So let's bring up the tab visualization that we just did in the last lesson. The first thing we'll do is right click on the scatter pot that we've been working on. There is a bunch of options here, but I want you to select duplicate sheet. And once that sheet is duplicated, rename it as discounted sales. Just something to identify it because you're going to be working with that one a bit. So make sure you're in that sheet, called discounted sales. Click on Analysis, then click on Aggregate Measures, that will check Aggregate Measures in that menu. Because the scatter plot is not aggregate, we need to re-aggregate the data, to be able to get a summary to test our hypotheses on discounts. Now all you see in front of you is a horizontal line, so I wanted you to just click the Show Me tab and then click on the cross tab just to get you some text. And we're going to be working with that for a few minutes. Field, so now we're going to create a calculated field called profitloss. And I'm just arbitrarily naming it profitloss, one word. Again, this is working towards looking at discounts as a reason why some stuff is selling at a loss. So the formula is as follows If the profit ratio is less than or equal to zero then the profit loss is equal to loss. Otherwise it's profit, so again if the profit loss is less than or equal to zero then profit loss is equal to loss otherwise it is profit. And now I'll change if it's not already that way, make sure. That profit loss is at discrete dimension, it should be a discrete dimension variable. The reason is that is because I want to see the differences in discounts between sales that we're at a loss in sales, that we're profitable. So let's drag the profit ratio and sales field away from the visualization. Now, move the discount field to measure values, the average discount given to the profitable sales was 8%. While those making a loss was discounted 47%, in other words selling stuff at big discounts tanks profits. Now, it's entirely possible that you know they wanted to get rid of it, they just wanted to clear out the warehouse for new stuffs. So the context is in important and we don't have that context but it's very interesting to see this really in action. So how can we show this in a way that taps into our iconic or at least short term memory and make it thus. Rather accessible and usable, I want to take a second here and put in a plug for boring old text. Text is awesome and thoughtfully deployed is essential to usability of your visualization. But the way I have the text here is a little bit hard to be able to interpret. It uses too much of brain power, so I'm going to make some changes to make it more accessible and add some texting to help out. So first I want to change the discount from a decimal to a percent, just because it's easier for us to understand. We're sort of used to seeing percentages at for discounts. So we're just going to change that now, then I'm going to use the power of Tableau here to generate a graph. And all I'm doing is dragging the sum of profit field from where it is now to the column shelf and it automatically does something interesting, it turns it into a graph. Now take the profitloss field from the left and drop it over color. The colors Tableau chooses might be different on your screen then on my screen. But we need to change them anyways, so I'm going to choose a light grey for profit and red for loss. I want to add some text context to it, remember I said a couple minutes ago that text is very important so I want to drag profitloss and profit to text. So now click on the label, click on the three dots you see in front of me to edit the label. Type in format text to read discount for summer profit of profitloss is average discount. And what that is, is just basically putting in the field there so that Tableau knows what number to add or a text to add, for the particular place that the label goes. I'm going to get rid of all the axis because we have data labels so we don't need an axis. However, before I do that, I'm going to add what's known as a reference line. And this reference line, is simply a way to show an illustration that I feel is necessary. So I'm going to put a zero reference line so that we know where the zero is 0%, or zero profit rather. And then to the left of it is going to be loss, to the right is profit, and so it's a pretty straight forward reference line. This may be the first time you've seen a reference line, and this is how you do it. You create a reference line here, and you put the information on the graph, and now I can remove the axis. So it's a nice way to actually show give you a little bit more context. Now I'm going to tweak the location of the text just a little bit. Your mileage might vary on how you want to tweak it but there you go, and it's done. So what we have done here is actually make a presentation that could be a little bit overwhelming to someone much more accessible and usable. And the key is that you as the data analyst are the ones who are able to analyze it and pick out the details that are needed to be able to show the decision makers. The information so that she can turn around and say okay, what are we going to do with this? What sort of actions if any are we going to take on this information. So I'll see you next time, take care, bye. Simply move forward with some sort of actionable piece here, so until next time, have a good day.
