9:59:59.000,9:59:59.000
1[br]00:00:04,540 --> 00:00:10,180[br]Welcome in this video, I'm going to start introducing the basic concepts of statistical graphics.

9:59:59.000,9:59:59.000
2[br]00:00:10,180 --> 00:00:13,660[br]I want you to be able to understand the value of graphics for presenting data,

9:59:59.000,9:59:59.000
3[br]00:00:13,660 --> 00:00:19,540[br]identify parts of a statistical image, and understand some pitfalls and graphics that we want to try to avoid.

9:59:59.000,9:59:59.000
4[br]00:00:19,540 --> 00:00:25,300[br]So here's an example of a chart, and there's a variety of different pieces of this chart.

9:59:59.000,9:59:59.000
5[br]00:00:25,300 --> 00:00:31,900[br]We have an x axis. That's the horizontal x axis.

9:59:59.000,9:59:59.000
6[br]00:00:31,900 --> 00:00:41,820[br]We have a y axis, the vertical axis. Each of these axes has a label.

9:59:59.000,9:59:59.000
7[br]00:00:41,820 --> 00:00:52,610[br]Task to task one. We have a caption up at the top that explains what's going on in the image, provides us with the context to understand it.

9:59:59.000,9:59:59.000
8[br]00:00:52,610 --> 00:00:59,660[br]And it says that this graph is showing the number of queries per task with query account distributions in the margins.

9:59:59.000,9:59:59.000
9[br]00:00:59,660 --> 00:01:04,070[br]And each dot is one participant. So it tells us we have a data point in it.

9:59:59.000,9:59:59.000
10[br]00:01:04,070 --> 00:01:09,020[br]What is that? It tells us what it is that we're charting. Number of queries per task.

9:59:59.000,9:59:59.000
11[br]00:01:09,020 --> 00:01:14,760[br]When we then see the axis labels that we have task one and we have task to.

9:59:59.000,9:59:59.000
12[br]00:01:14,760 --> 00:01:19,500[br]Those two together, give us the context, understand that. Oh, we have two tasks.

9:59:59.000,9:59:59.000
13[br]00:01:19,500 --> 00:01:24,000[br]And this is why of one participant.

9:59:59.000,9:59:59.000
14[br]00:01:24,000 --> 00:01:28,620[br]And they're appearing at the point where they have their task, one count on their task, two counts.

9:59:59.000,9:59:59.000
15[br]00:01:28,620 --> 00:01:34,530[br]OK, this allows us to to see if there's any relationship between how long it took to.

9:59:59.000,9:59:59.000
16[br]00:01:34,530 --> 00:01:40,800[br]How many queries it took to complete the two different tasks. It then says we have query count distribution than the margin.

9:59:59.000,9:59:59.000
17[br]00:01:40,800 --> 00:01:46,050[br]So this is a compound plot. And in the left and right, margins are in the X and Y margins.

9:59:59.000,9:59:59.000
18[br]00:01:46,050 --> 00:01:52,530[br]We have the distribution, a histogram of the X axis, the task one.

9:59:59.000,9:59:59.000
19[br]00:01:52,530 --> 00:01:59,560[br]We have a histogram of the Y axis task to these histograms don't have axes themselves because,

9:59:59.000,9:59:59.000
20[br]00:01:59,560 --> 00:02:04,680[br]well, we just wanted to show a distribution the exact particularly for our purposes here.

9:59:59.000,9:59:59.000
21[br]00:02:04,680 --> 00:02:16,980[br]The exact number in each bean is not so important. The key thing is just see being able to see relatively where is the mass of the different?

9:59:59.000,9:59:59.000
22[br]00:02:16,980 --> 00:02:21,150[br]Where is the mass on the two different task counts?

9:59:59.000,9:59:59.000
23[br]00:02:21,150 --> 00:02:27,620[br]And we can see that both of them have a a right skew.

9:59:59.000,9:59:59.000
24[br]00:02:27,620 --> 00:02:32,780[br]They're bulked up towards towards the low end of the scale.

9:59:59.000,9:59:59.000
25[br]00:02:32,780 --> 00:02:41,120[br]And then we have the all the we have all of the individual data points scat on the chart.

9:59:59.000,9:59:59.000
26[br]00:02:41,120 --> 00:02:46,670[br]This is called a scatterplot. We have these different pieces of the chart that we want to be able to identify.

9:59:59.000,9:59:59.000
27[br]00:02:46,670 --> 00:02:55,400[br]And when you go particularly as you go to refine a chart, what you're going to need to do is specify what's happening on each of these pieces.

9:59:59.000,9:59:59.000
28[br]00:02:55,400 --> 00:02:59,300[br]What is your x axis? What is your y axis?

9:59:59.000,9:59:59.000
29[br]00:02:59,300 --> 00:03:08,600[br]Before you even start the chart, you need to set up your data so that we have what is the data point that I'm going to be plotting on this chart?

9:59:59.000,9:59:59.000
30[br]00:03:08,600 --> 00:03:16,280[br]So charts can are really useful for revealing a variety of things that can reveal patterns or lack there of.

9:59:59.000,9:59:59.000
31[br]00:03:16,280 --> 00:03:19,200[br]In this chart, there's really not much of a pattern.

9:59:59.000,9:59:59.000
32[br]00:03:19,200 --> 00:03:26,700[br]And we can see that it's booked up, but particularly if we get out to that larger number of tasks, there's not a lot of pattern.

9:59:59.000,9:59:59.000
33[br]00:03:26,700 --> 00:03:34,560[br]The one with the participant, with the most tasks and with the most tasks or queries for task one.

9:59:59.000,9:59:59.000
34[br]00:03:34,560 --> 00:03:38,760[br]Has a middling to low number of queries for Task two.

9:59:59.000,9:59:59.000
35[br]00:03:38,760 --> 00:03:47,790[br]And the the one who has the most queries for task to while they're in the upper end of of the queries per task on task one, they're biased.

9:59:59.000,9:59:59.000
36[br]00:03:47,790 --> 00:03:53,520[br]They're not at all the highest. So we can see there's not a not very much of a relationship here.

9:59:59.000,9:59:59.000
37[br]00:03:53,520 --> 00:03:59,490[br]At least that doesn't look like one. They can be useful for comparisons. If we've got a bar chart, we can compare to bars.

9:59:59.000,9:59:59.000
38[br]00:03:59,490 --> 00:04:08,220[br]We can see where points lay. We can see like we can see in that chart that we just saw that the the highest number of counts for one task,

9:59:59.000,9:59:59.000
39[br]00:04:08,220 --> 00:04:12,210[br]the highest number of counts for another task or different. We can also see trends.

9:59:59.000,9:59:59.000
40[br]00:04:12,210 --> 00:04:14,610[br]We can see if a line looks like it goes up or down,

9:59:59.000,9:59:59.000
41[br]00:04:14,610 --> 00:04:22,950[br]wiggles around so they can reveal a lot of these kinds of things and they can really leverage our human perception and our human,

9:59:59.000,9:59:59.000
42[br]00:04:22,950 --> 00:04:25,080[br]particularly our human visual senses,

9:59:59.000,9:59:59.000
43[br]00:04:25,080 --> 00:04:35,820[br]to be able to quickly internalize and understand what is going on in in a set of data when we're creating a chart.

9:59:59.000,9:59:59.000
44[br]00:04:35,820 --> 00:04:41,520[br]We need to clearly document a few things. We you clearly state what is being presented when someone looks at a chart.

9:59:59.000,9:59:59.000
45[br]00:04:41,520 --> 00:04:45,720[br]They need to be able to understand what each point in the chart is going to be.

9:59:59.000,9:59:59.000
46[br]00:04:45,720 --> 00:04:56,530[br]They need to understand what values are plotted on the axis. They need to understand what values are plotted on the axes.

9:59:59.000,9:59:59.000
47[br]00:04:56,530 --> 00:05:01,260[br]Often this is done in an axis label in our in the chart I showed you,

9:59:59.000,9:59:59.000
48[br]00:05:01,260 --> 00:05:06,040[br]it said the values in the caption in the axis labels said which version of them they were.

9:59:59.000,9:59:59.000
49[br]00:05:06,040 --> 00:05:08,560[br]If there are units, that needs to be clear.

9:59:59.000,9:59:59.000
50[br]00:05:08,560 --> 00:05:19,840[br]So if you've got something that's millimeters, that's pounds, that's megabytes, whatever, you need to specify the units in your in your chart,

9:59:59.000,9:59:59.000
51[br]00:05:19,840 --> 00:05:29,320[br]either in the Axis label or in the caption, some of these things can sometimes be implicit in the type of chart, such as a histogram.

9:59:59.000,9:59:59.000
52[br]00:05:29,320 --> 00:05:32,440[br]And you've got a fraction or a percentage in the left hand side.

9:59:59.000,9:59:59.000
53[br]00:05:32,440 --> 00:05:41,260[br]It's standard convention that we're talking about, the fraction of the values that are in each bin,

9:59:59.000,9:59:59.000
54[br]00:05:41,260 --> 00:05:49,930[br]at least if you label it as a histogram or as a chart showing the distribution. But when in doubt, if there's any doubt about.

9:59:59.000,9:59:59.000
55[br]00:05:49,930 --> 00:05:57,090[br]What a value, what an axis label is. Or there's any doubt that the reader will understand what it is.

9:59:59.000,9:59:59.000
56[br]00:05:57,090 --> 00:06:04,230[br]Be explicit, explicitly, say what's going on in your chart. That also the chart in the caption should be interpretable on their own.

9:59:59.000,9:59:59.000
57[br]00:06:04,230 --> 00:06:08,640[br]You can assume a reasonable level. You have to know your audience for this.

9:59:59.000,9:59:59.000
58[br]00:06:08,640 --> 00:06:16,200[br]But someone should be able to just look at the chart with its immediately surrounding description,

9:59:59.000,9:59:59.000
59[br]00:06:16,200 --> 00:06:20,670[br]the labels, the caption and understand have a pretty good idea of what's going on.

9:59:59.000,9:59:59.000
60[br]00:06:20,670 --> 00:06:25,990[br]The surrounding text with the text that references the chart if you're writing a document.

9:59:59.000,9:59:59.000
61[br]00:06:25,990 --> 00:06:30,190[br]That can have your observations, that can provide more context and clarity.

9:59:59.000,9:59:59.000
62[br]00:06:30,190 --> 00:06:35,950[br]But someone just looking at the chart should be able to figure out basically what's going on and

9:59:59.000,9:59:59.000
63[br]00:06:35,950 --> 00:06:41,080[br]not be too far off of this is particularly important because there's a there's a lot of people,

9:59:59.000,9:59:59.000
64[br]00:06:41,080 --> 00:06:46,150[br]whether this is a good or a bad practice, we can debate. But there's a lot of people who, when they're reading a paper,

9:59:59.000,9:59:59.000
65[br]00:06:46,150 --> 00:06:52,000[br]they focus on the charts and look at the key charts first to see what it is that's going on in the paper.

9:59:59.000,9:59:59.000
66[br]00:06:52,000 --> 00:07:01,990[br]And if our if our charts are self-explanatory and are clear that it makes a lot easier for people to glance at our work,

9:59:59.000,9:59:59.000
67[br]00:07:01,990 --> 00:07:09,370[br]see what it's doing and decide whether they are going to pay it further attention.

9:59:59.000,9:59:59.000
68[br]00:07:09,370 --> 00:07:12,970[br]So in a paper, if you're putting a chart in a docket, a written document or a paper,

9:59:59.000,9:59:59.000
69[br]00:07:12,970 --> 00:07:18,640[br]each figure should have a caption and the caption can it labels the figure and it can also provide interpretive guidance.

9:59:59.000,9:59:59.000
70[br]00:07:18,640 --> 00:07:18,790[br]Like,

9:59:59.000,9:59:59.000
71[br]00:07:18,790 --> 00:07:26,410[br]it's not uncommon for a caption to be two or three sentences saying things about what's going on in the chart and describing some of the methodology,

9:59:59.000,9:59:59.000
72[br]00:07:26,410 --> 00:07:31,800[br]what precisely some of the computations are, etc. In other contexts, we often need a title for the charts.

9:59:59.000,9:59:59.000
73[br]00:07:31,800 --> 00:07:37,900[br]So if we have a caption we don't, we need to label our axes, but we don't need a title for the chart itself all the time.

9:59:59.000,9:59:59.000
74[br]00:07:37,900 --> 00:07:40,840[br]It doesn't hurt, but often it's redundant with the caption.

9:59:59.000,9:59:59.000
75[br]00:07:40,840 --> 00:07:46,690[br]In other contexts, though, we often do need a title such as when we have a chart that's going in a presentation.

9:59:59.000,9:59:59.000
76[br]00:07:46,690 --> 00:07:50,230[br]We have a chart in one of our notebooks. A title is often helpful in notebooks.

9:59:59.000,9:59:59.000
77[br]00:07:50,230 --> 00:07:52,180[br]The surrounding text may be sufficient,

9:59:59.000,9:59:59.000
78[br]00:07:52,180 --> 00:07:59,920[br]but a title is often a good idea for someone who's quickly scanning the notebook to be able to understand what's going on in the chart.

9:59:59.000,9:59:59.000
79[br]00:07:59,920 --> 00:08:04,570[br]So a few pitfalls to be aware of when we're thinking about statistical graphics is

9:59:59.000,9:59:59.000
80[br]00:08:04,570 --> 00:08:10,930[br]one is distorting the distances or the differences that are happening particularly.

9:59:59.000,9:59:59.000
81[br]00:08:10,930 --> 00:08:13,600[br]We need to make sure if something has a length,

9:59:59.000,9:59:59.000
82[br]00:08:13,600 --> 00:08:21,040[br]anything that has a length that length should accurately represent quantities, position, relative position.

9:59:59.000,9:59:59.000
83[br]00:08:21,040 --> 00:08:26,800[br]If you have two dots, their relative position is what's important. But if we have a length of it, if we have a bar, it has a length.

9:59:59.000,9:59:59.000
84[br]00:08:26,800 --> 00:08:30,700[br]It also is an area we need to make sure those accurately represent quantities.

9:59:59.000,9:59:59.000
85[br]00:08:30,700 --> 00:08:37,140[br]One really common way to violate this is having a bar chart whose access starts at something other than zero.

9:59:59.000,9:59:59.000
86[br]00:08:37,140 --> 00:08:41,670[br]The software we're using doesn't do that by default. Excel does.

9:59:59.000,9:59:59.000
87[br]00:08:41,670 --> 00:08:46,860[br]But your bar chart always needs to start at zero because people are beat.

9:59:59.000,9:59:59.000
88[br]00:08:46,860 --> 00:08:51,540[br]People don't look at the relative position of the bar. People see the whole height of the bar.

9:59:59.000,9:59:59.000
89[br]00:08:51,540 --> 00:08:59,400[br]And so if it doesn't start at zero, it looks like the difference between bars is much higher relative to the bar size than it actually is.

9:59:59.000,9:59:59.000
90[br]00:08:59,400 --> 00:09:01,860[br]There's also ways in which we can violate conventions.

9:59:59.000,9:59:59.000
91[br]00:09:01,860 --> 00:09:09,210[br]So in the first video I showed you the chart that violated the convention, that the x axis goes in order.

9:59:59.000,9:59:59.000
92[br]00:09:09,210 --> 00:09:16,320[br]If we violate the user's expectations, they they'll either be confused by the chart or read it wrong.

9:59:59.000,9:59:59.000
93[br]00:09:16,320 --> 00:09:24,240[br]Statistical graphics in each particular type of chart have conventions that people who read a lot of them assimilate by long patterns of reading,

9:59:59.000,9:59:59.000
94[br]00:09:24,240 --> 00:09:32,570[br]like you assimilate how to read written text. And if those expect expectations are violated, that can.

9:59:59.000,9:59:59.000
95[br]00:09:32,570 --> 00:09:37,490[br]Lead the user to incorrect conclusions from our charts, from our presentation.

9:59:59.000,9:59:59.000
96[br]00:09:37,490 --> 00:09:42,920[br]A key thing to remember here that also applies to all of our presentations.

9:59:59.000,9:59:59.000
97[br]00:09:42,920 --> 00:09:52,350[br]Research isn't a mystery novel. You don't have to worry about spoiling the surprise or you end the goal here is not to present it,

9:59:59.000,9:59:59.000
98[br]00:09:52,350 --> 00:09:56,100[br]not to subvert tropes or present shocking new presentations.

9:59:59.000,9:59:59.000
99[br]00:09:56,100 --> 00:10:00,470[br]We might have shocking new evidence, but from a presentation perspective,

9:59:59.000,9:59:59.000
100[br]00:10:00,470 --> 00:10:08,040[br]we want it to fit within conventions and not violate readers expectations unnecessarily so that

9:59:59.000,9:59:59.000
101[br]00:10:08,040 --> 00:10:14,370[br]they can read it and be confident that they've correctly understood what it is that you're saying.

9:59:59.000,9:59:59.000
102[br]00:10:14,370 --> 00:10:18,990[br]Another thing to be aware of is that graphics can illustrate an effect.

9:59:59.000,9:59:59.000
103[br]00:10:18,990 --> 00:10:27,060[br]They can also help you find an effect. Like more exploring data. We can look at the graphics to see what effects we might be looking for.

9:59:59.000,9:59:59.000
104[br]00:10:27,060 --> 00:10:33,660[br]We have to be careful about that. We'll talk about some of the pitfalls of we have to be careful, more combining.

9:59:59.000,9:59:59.000
105[br]00:10:33,660 --> 00:10:39,420[br]We can't combine exploratory and what's called confirmatory analysis, but they can help us.

9:59:59.000,9:59:59.000
106[br]00:10:39,420 --> 00:10:45,990[br]Visualizing data can help us look for possible effects and get ideas for what to go look for next.

9:59:59.000,9:59:59.000
107[br]00:10:45,990 --> 00:10:53,530[br]But they're not conclusive proof of an effect. We need the numeric results, just the raw numerical.

9:59:59.000,9:59:59.000
108[br]00:10:53,530 --> 00:11:00,030[br]The raw numbers as well as the numeric result are the results of inferential techniques that let us

9:59:59.000,9:59:59.000
109[br]00:11:00,030 --> 00:11:06,090[br]estimate how big an effect is and whether it's significant in order to come to any conclusions.

9:59:59.000,9:59:59.000
110[br]00:11:06,090 --> 00:11:13,920[br]So in this chart, I want to show you, for example, we have if we look at the chart closely,

9:59:59.000,9:59:59.000
111[br]00:11:13,920 --> 00:11:20,790[br]we have these two data points and the little blue Xs in them, the func SVOD axis.

9:59:59.000,9:59:59.000
112[br]00:11:20,790 --> 00:11:26,160[br]As to the left of the item, item X. So it looks like for this particular metric,

9:59:59.000,9:59:59.000
113[br]00:11:26,160 --> 00:11:32,900[br]lower is a better value for it because it's an error metric root mean squared errors with RMX he stands for.

9:59:59.000,9:59:59.000
114[br]00:11:32,900 --> 00:11:38,520[br]But it looks like. Okay. This is a little bit better, but that's not sufficient evidence for us to include.

9:59:59.000,9:59:59.000
115[br]00:11:38,520 --> 00:11:44,220[br]To conclude that func SVOD is better than item item on the per user are masc metric.

9:59:59.000,9:59:59.000
116[br]00:11:44,220 --> 00:11:48,900[br]Exactly what all these things are is, is a topic for another day.

9:59:59.000,9:59:59.000
117[br]00:11:48,900 --> 00:11:55,230[br]But the fact that we see the thing to the left, that s that if the effect is real, this illustrates it.

9:59:59.000,9:59:59.000
118[br]00:11:55,230 --> 00:12:02,220[br]But seeing it's not enough for us to conclude that it outperforms because it might be a fluke of our experimental strategy,

9:59:59.000,9:59:59.000
119[br]00:12:02,220 --> 00:12:07,120[br]it's a relatively small difference. So. They help us see.

9:59:59.000,9:59:59.000
120[br]00:12:07,120 --> 00:12:11,620[br]They help us communicate. They're not definitive and conclusive proof.

9:59:59.000,9:59:59.000
121[br]00:12:11,620 --> 00:12:16,660[br]Couple of other things I want to highlight. They're going on in this graph. I've introduced two different kinds of symbols here.

9:59:59.000,9:59:59.000
122[br]00:12:16,660 --> 00:12:19,930[br]So the earlier graph, we just had one kind of symbol. We had dots here.

9:59:59.000,9:59:59.000
123[br]00:12:19,930 --> 00:12:24,580[br]We have two different kinds with a legend that says red circles are global.

9:59:59.000,9:59:59.000
124[br]00:12:24,580 --> 00:12:30,930[br]Oremus are a thing called global are masc and blue Xs are a thing called per user are MSE.

9:59:59.000,9:59:59.000
125[br]00:12:30,930 --> 00:12:37,920[br]Don't have to understand what those are. But the point is, I'm using different colors and shapes in order to communicate,

9:59:59.000,9:59:59.000
126[br]00:12:37,920 --> 00:12:45,750[br]to show different versions of a thing in the same chart, using different shapes.

9:59:59.000,9:59:59.000
127[br]00:12:45,750 --> 00:12:50,430[br]In addition to different colors is useful because it's so imprinted on a black and white printer.

9:59:59.000,9:59:59.000
128[br]00:12:50,430 --> 00:12:57,270[br]If they if they have some form of color blindness, it helps it make the differences clearer.

9:59:59.000,9:59:59.000
129[br]00:12:57,270 --> 00:13:05,070[br]I've also in addition so I've got my Y at my at Y axis, which is indicating different things that I'm plotting here.

9:59:59.000,9:59:59.000
130[br]00:13:05,070 --> 00:13:11,430[br]I also have grouped them just to make it easier for the user to see.

9:59:59.000,9:59:59.000
131[br]00:13:11,430 --> 00:13:15,330[br]These are the same like these. These first ones are all single algorithms.

9:59:59.000,9:59:59.000
132[br]00:13:15,330 --> 00:13:20,920[br]And then we have a blend and a few other things. The details are.

9:59:59.000,9:59:59.000
133[br]00:13:20,920 --> 00:13:27,220[br]Aren't important for illustrating them, but it helps guide the user to understand his structures to we have these group breakdowns.

9:59:59.000,9:59:59.000
134[br]00:13:27,220 --> 00:13:32,680[br]It also helps save space in the paper because I can present all of these different things in one place.

9:59:59.000,9:59:59.000
135[br]00:13:32,680 --> 00:13:39,730[br]It's easy to compare the different stages, even though I have to split the mountain, the discussion in the paper.

9:59:59.000,9:59:59.000
136[br]00:13:39,730 --> 00:13:46,540[br]But it gives you one place to compare them and it concisely shows the key results of the entire paper in one chart.

9:59:59.000,9:59:59.000
137[br]00:13:46,540 --> 00:13:51,880[br]So to wrap up graphics can make data clearer and they let us leverage human perception to understand it.

9:59:59.000,9:59:59.000
138[br]00:13:51,880 --> 00:13:54,160[br]They don't replace our numerical analysis,

9:59:59.000,9:59:59.000
139[br]00:13:54,160 --> 00:14:01,570[br]but they give it context and they help us more clearly communicate what it is that we're learning from the data and what's going on in it.

9:59:59.000,9:59:59.000
140[br]00:14:01,570 --> 00:14:06,340[br]We do, however, always need to make sure that we clearly label and describe our graphics so that

9:59:59.000,9:59:59.000
141[br]00:14:06,340 --> 00:14:15,300[br]readers can understand them and they can draw correct conclusions from them.

9:59:59.000,9:59:59.000