1
00:00:05,955 --> 00:00:07,057
[MUSIC]. 
Hi Everyone. 

2
00:00:07,057 --> 00:00:11,457
My name is Cecilia Aragon and I'll be 
your guest lecturer this week. 

3
00:00:11,457 --> 00:00:15,717
We're going to be talking about basic 
principles of information visualization. 

4
00:00:15,717 --> 00:00:20,397
This is in my opinion one of the most 
important interesting and fun components 

5
00:00:20,397 --> 00:00:25,149
of the data science pipeline. 
We're going to be talking about how you 

6
00:00:25,149 --> 00:00:28,549
take the data that you've collected and 
analyzed, and then actually present it in 

7
00:00:28,549 --> 00:00:31,881
a form that makes sense to the human 
brain. 

8
00:00:31,881 --> 00:00:35,511
I'm a professor here at the University of 
Washington in the Department of Human 

9
00:00:35,511 --> 00:00:40,498
Centered Design and Engineering. 
And it'll be my pleasure to introduce you 

10
00:00:40,498 --> 00:00:45,736
to this exciting subject. 
My research interests lie in the visual 

11
00:00:45,736 --> 00:00:51,521
presentation of very large, scientific 
and social datasets. 

12
00:00:51,521 --> 00:00:55,501
So, let's start with a few definitions. 
The word visualization is extremely 

13
00:00:55,501 --> 00:00:58,384
overloaded. 
There are many possible meanings to the 

14
00:00:58,384 --> 00:01:01,684
term so what are we talking about when we 
say we're studying Information 

15
00:01:01,684 --> 00:01:06,310
Visualization? 
Well we want to consider the visual 

16
00:01:06,310 --> 00:01:11,800
representation of information. 
The goals of Information Visualization 

17
00:01:11,800 --> 00:01:15,820
include the effective communication of 
information or data, with clarity that 

18
00:01:15,820 --> 00:01:19,300
means that the information should be 
presented in a way that's easy to 

19
00:01:19,300 --> 00:01:26,040
understand by a human. 
With integrity, you present all the data 

20
00:01:26,040 --> 00:01:30,790
and nothing but the data. 
And you want to stimulate viewer 

21
00:01:30,790 --> 00:01:34,590
engagement with the data. 
Our focus in this course will be on 

22
00:01:34,590 --> 00:01:37,850
effectiveness. 
We're talking about the clear and 

23
00:01:37,850 --> 00:01:40,780
effective presentation of data to the 
human mind. 

24
00:01:42,720 --> 00:01:48,180
Data visualization usually falls into 
three academic categories. 

25
00:01:48,180 --> 00:01:53,246
Information visualization, scientific 
visualization and visual analytics. 

26
00:01:53,246 --> 00:02:00,510
Information visualization typically 
involves abstract information that may 

27
00:02:00,510 --> 00:02:04,670
not have a position in space. 
For examlpe, on the slide you can see a 

28
00:02:04,670 --> 00:02:07,946
graph of stock prices around the year 
2000. 

29
00:02:07,946 --> 00:02:14,350
We graphed the price, the dollar amount 
on the y-axis and time on the x axis. 

30
00:02:14,350 --> 00:02:18,860
Scientific visualization usually involves 
displaying data that has a physical 

31
00:02:18,860 --> 00:02:22,340
position in space. 
For example, here you see a methane 

32
00:02:22,340 --> 00:02:26,540
combustion data set, where each of the 
data points has a 3D location. 

33
00:02:27,960 --> 00:02:33,700
Visual analytics is a combination of 
highly interactive visual interfaces and 

34
00:02:33,700 --> 00:02:36,820
sophisticated statistical learning 
algorithms. 

35
00:02:36,820 --> 00:02:40,605
In this course we're going to be talking 
about basic principles of information 

36
00:02:40,605 --> 00:02:43,360
visualizations. 
We're not going to be talking about 

37
00:02:43,360 --> 00:02:48,970
scientific visualization specifically, 
nor about visual analytics, nor about the 

38
00:02:48,970 --> 00:02:54,420
artistic field of visual design. 
The principles, however, of information 

39
00:02:54,420 --> 00:02:58,680
visualization that we are discussing are 
applicable to all three domains. 

40
00:02:59,890 --> 00:03:05,060
We will be discussing specifically this 
week how to map data variables to visual 

41
00:03:05,060 --> 00:03:09,110
attributes. 
So, why is visualization important? 

42
00:03:09,110 --> 00:03:12,870
Say you have a million numbers. 
What's the best way to communicate that 

43
00:03:12,870 --> 00:03:18,450
information to the human brain? 
Well, imagine a table of a million 

44
00:03:18,450 --> 00:03:23,960
numbers, it seems almost too vast to 
comprehend, much less define meaningful 

45
00:03:23,960 --> 00:03:28,550
trends or information about it. 
But on the other hand, if you consider 

46
00:03:28,550 --> 00:03:32,950
the number of pixels on your computer 
screen it's typically around a million 

47
00:03:32,950 --> 00:03:35,774
and you have no trouble understanding it 
very quickly. 

48
00:03:35,774 --> 00:03:41,120
And being able to, to act on the 
information you [UNKNOWN], so the human 

49
00:03:41,120 --> 00:03:44,322
visual system is the highest-bandwith 
channel to the human brain. 

50
00:03:44,322 --> 00:03:47,900
And it's the most efficient way to 
present large amounts of information so 

51
00:03:47,900 --> 00:03:54,260
they make sense. 
Lets talk about a few examples, so may be 

52
00:03:54,260 --> 00:03:58,610
the first use of visualization 
historically was for eliciting knowledge 

53
00:03:58,610 --> 00:04:03,680
from data, let's suppose you have a table 
of numbers. 

54
00:04:03,680 --> 00:04:07,440
We're, we'll start with a relatively 
small table with only 50 rows, one for 

55
00:04:07,440 --> 00:04:10,970
each state in the union, three columns in 
each row. 

56
00:04:10,970 --> 00:04:15,480
The name of the state, the percentage of 
college graduates in that state, and the 

57
00:04:15,480 --> 00:04:19,410
per capita income. 
So you might have questions about this 

58
00:04:19,410 --> 00:04:22,720
data, such as, which state has the 
highest income? 

59
00:04:22,720 --> 00:04:26,240
Is there a relationship between income 
and education? 

60
00:04:26,240 --> 00:04:28,520
Finally are there any outliers? 
Are there any trends? 

61
00:04:30,140 --> 00:04:34,130
Well obviously all the data is present, 
and you could take your time looking at 

62
00:04:34,130 --> 00:04:37,170
this data. 
You could identify the highest and lowest 

63
00:04:37,170 --> 00:04:39,240
numbers. 
You could answer all these questions from 

64
00:04:39,240 --> 00:04:41,700
the data. 
Everything is presented here. 

65
00:04:41,700 --> 00:04:44,830
But is this the most efficient way to 
understand this data? 

66
00:04:46,970 --> 00:04:50,030
Here's another way of presenting the data 
in a scatter plot. 

67
00:04:50,030 --> 00:04:53,620
We have per capita income on the x axis 
and the percentage of college degrees on 

68
00:04:53,620 --> 00:04:57,540
the y axis. 
And, as you can see, the question of 

69
00:04:57,540 --> 00:05:00,500
which state has the highest income is 
instantly answered. 

70
00:05:00,500 --> 00:05:03,970
We can see that Connecticut has the 
highest per capita income. 

71
00:05:03,970 --> 00:05:08,800
Similarly, it's clear there's a trend 
that there's a correlation between per 

72
00:05:08,800 --> 00:05:12,100
capita income and the percentage of 
college of degrees and there is a couple 

73
00:05:12,100 --> 00:05:17,070
of interesting outliers. 
Nevada has a relatively high per capita 

74
00:05:17,070 --> 00:05:20,270
income but a very low percentage of 
college graduates. 

75
00:05:20,270 --> 00:05:25,250
So, all the information is presented in 
both of these forms but the, the visual 

76
00:05:25,250 --> 00:05:32,130
representation is, is much more efficient 
at presenting meaning to the human mind. 

77
00:05:35,770 --> 00:05:39,520
Well, so, okay, the first question that 
might come up is, well, yeah, but all 

78
00:05:39,520 --> 00:05:43,850
this information could more easily have 
been accomplished by analyzing the data 

79
00:05:43,850 --> 00:05:46,640
with a statistical program. 
You could find the maximum. 

80
00:05:46,640 --> 00:05:50,120
You can find the minimum. 
You can find the regression line. 

81
00:05:50,120 --> 00:05:55,130
However, it turns out the human brain is 
one of the best pattern matching tools 

82
00:05:55,130 --> 00:05:58,040
that we have. 
Actually, it's, we're, the human brain's 

83
00:05:58,040 --> 00:06:03,650
so far is still better at detecting 
patterns in data than any of our most 

84
00:06:03,650 --> 00:06:05,570
sophisticated machine learning 
algorithms. 

85
00:06:06,900 --> 00:06:11,870
So an immediate question that arises is. 
Why not just use a statistical program 

86
00:06:11,870 --> 00:06:16,530
and identify the minimum, the maximum, 
the trend lines and go from there? 

87
00:06:16,530 --> 00:06:23,820
Why bother with a visual representation? 
The answer is simply, the graphs reveal 

88
00:06:23,820 --> 00:06:29,850
data that statistics may not. 
So, here's a famous example, Anscombe's 

89
00:06:29,850 --> 00:06:35,150
Quartet. 
Here you have four small data sets that 

90
00:06:35,150 --> 00:06:40,180
are quite different, as you can see by 
looking at them, but they have identical 

91
00:06:40,180 --> 00:06:42,750
linear models. 
That is, the identical means and the 

92
00:06:42,750 --> 00:06:46,510
identical trend lines. 
Again, you could just present the data 

93
00:06:46,510 --> 00:06:52,110
in, in tabular form, but see how much 
more information we can learn about 

94
00:06:52,110 --> 00:06:55,810
these, these individual data sets, just 
by looking at the data. 

95
00:06:55,810 --> 00:07:00,145
So even though, we have identical means, 
identical regression lines. 

96
00:07:00,145 --> 00:07:04,760
These data sets are quite different. 
So, so statistics may not tell us 

97
00:07:04,760 --> 00:07:11,380
everything about a particular data set. 
I hope I've managed to wet your appetite 

98
00:07:11,380 --> 00:07:15,750
for this coming weeks set of slides. 
I'm going to stop here and we'll continue 

99
00:07:15,750 --> 00:07:17,770
on with more examples in the next 
lecture.