1 00:00:03,230 --> 00:00:06,240 Chapter 2 has very simple thesis. 2 00:00:07,280 --> 00:00:10,340 And it comes from Jock Mackinlay who is one of 3 00:00:10,340 --> 00:00:14,890 the I think the trailblazers in modern computational visualization. 4 00:00:14,890 --> 00:00:18,440 And he came up with this idea that he called Effectiveness. 5 00:00:18,440 --> 00:00:19,510 And he basically says, 6 00:00:19,510 --> 00:00:24,410 look, how can we tell if one visualization is better than another one? 7 00:00:24,410 --> 00:00:29,420 And he just came up with this heuristic, but I think it's very effective. 8 00:00:29,420 --> 00:00:31,920 And it relies on our understanding of human perception. 9 00:00:32,990 --> 00:00:34,870 So you can see he says a visualization, it, 10 00:00:34,870 --> 00:00:40,390 it's just more effective than another one, if the information it's conveyed is more 11 00:00:40,390 --> 00:00:46,570 readily perceived than the information in the other visualization. 12 00:00:47,720 --> 00:00:53,120 So in other words, in, if you want to learn how to make visualizations that can 13 00:00:53,120 --> 00:00:59,810 communicate powerfully, it's critical to understand how the human mind 14 00:00:59,810 --> 00:01:06,700 perceives shape, color, line, space, motion, and interaction. 15 00:01:06,700 --> 00:01:12,578 And these elements together, shape how we actually perceive what's on the display. 16 00:01:12,578 --> 00:01:17,770 So in this section, I'm going to quickly talk about 17 00:01:17,770 --> 00:01:24,170 some of the important qualities of human perception and see how they affect 18 00:01:24,170 --> 00:01:29,612 the way that we should structure data when we present it visually. 19 00:01:29,612 --> 00:01:35,200 So the first is just a quick walk through of Gestalt's ecology, which is basically 20 00:01:35,200 --> 00:01:39,510 a school of psychology that came out of Germany, and it really describes something 21 00:01:39,510 --> 00:01:45,970 very interesting, which is that the mind looks at objects in their entirety, 22 00:01:45,970 --> 00:01:50,290 before or in parallel with perception of individual parts. 23 00:01:51,350 --> 00:01:52,470 And this means that, 24 00:01:52,470 --> 00:01:55,480 you know, there's this very intriguing statement at the end here. 25 00:01:55,480 --> 00:01:59,590 The whole is other than the sum of its parts. 26 00:01:59,590 --> 00:02:01,610 So, let's talk about what this means. 27 00:02:02,790 --> 00:02:05,290 Look at this first concept here. 28 00:02:05,290 --> 00:02:07,640 The idea is called similarity. 29 00:02:07,640 --> 00:02:12,410 And what it means here, is that the mind takes shapes 30 00:02:12,410 --> 00:02:16,780 that have similar semantic relationship. 31 00:02:16,780 --> 00:02:20,240 Shape, and it brings them together. 32 00:02:20,240 --> 00:02:23,070 So what you see here is this eagle shape, but 33 00:02:23,070 --> 00:02:27,270 around the eagle are really effectively just a series of triangles. 34 00:02:28,470 --> 00:02:36,180 But our mind reads that those triangles create a kind of halo or circle behind 35 00:02:36,180 --> 00:02:42,250 the form of the eagle, if you look at the shapes, just by themselves. 36 00:02:42,250 --> 00:02:44,050 So don't think of them as a halo. 37 00:02:44,050 --> 00:02:47,080 What they are is just a series of triangles, but 38 00:02:47,080 --> 00:02:50,330 the mind reads them differently. 39 00:02:50,330 --> 00:02:54,830 So what's important to understand is how the mind reads these symbols together. 40 00:02:56,340 --> 00:02:58,240 I'll give you another example here. 41 00:02:58,240 --> 00:03:00,009 The principle is called Anomaly. 42 00:03:01,010 --> 00:03:06,900 An anomaly, very simply, tells the story that if there's continuity, and 43 00:03:06,900 --> 00:03:10,300 a break in the continuity, that that shape, 44 00:03:10,300 --> 00:03:14,650 that is no longer contiguous is going to stand out more. 45 00:03:15,780 --> 00:03:21,074 So what's important in human perception, is putting shapes together and 46 00:03:21,074 --> 00:03:25,370 identifying shapes that stand out from that togetherness. 47 00:03:28,007 --> 00:03:30,551 This has several interesting consequences. 48 00:03:30,551 --> 00:03:34,340 The first, is this principle called Continuity. 49 00:03:34,340 --> 00:03:41,130 And what this means, is that we the human mind will look at the cross, 50 00:03:41,130 --> 00:03:47,510 the swoop that goes across that H and read that as a continuous path. 51 00:03:47,510 --> 00:03:52,600 And while it's just a shape, it's, it's a negative space. 52 00:03:52,600 --> 00:03:57,950 What we really read that to mean, is that the leaf has blown through the H. 53 00:03:59,250 --> 00:04:02,720 So there's meaning in the con, the connection and 54 00:04:02,720 --> 00:04:05,910 continuation of shapes that are next to one another. 55 00:04:07,950 --> 00:04:11,850 Another important idea is something called Closure. 56 00:04:11,850 --> 00:04:13,690 And what this means, is that the mind 57 00:04:14,910 --> 00:04:21,190 makes shapes contiguous that are not formed entirely. 58 00:04:21,190 --> 00:04:27,260 And so if you look at the top of this panda bear shape, what you'll see is 59 00:04:27,260 --> 00:04:34,339 there's actually no top of the bear drawn, but your mind draws it in for you. 60 00:04:35,340 --> 00:04:42,320 So you can see how clearly the human mind is really interesting at forming patterns. 61 00:04:42,320 --> 00:04:47,740 And these patterns are not necessarily visible. 62 00:04:47,740 --> 00:04:51,080 So it's critical that we understand these patterns when we're 63 00:04:51,080 --> 00:04:54,450 putting together things that are more data oriented. 64 00:04:56,100 --> 00:04:59,960 So this other concept is called Proximity. 65 00:04:59,960 --> 00:05:06,510 And Proximity means that the mind believes that things that are closer together, 66 00:05:06,510 --> 00:05:08,390 have more meaning. 67 00:05:08,390 --> 00:05:14,440 In other words, the dots, the squares in this particular form, 68 00:05:14,440 --> 00:05:19,490 are less related than the forms in this particular shape. 69 00:05:20,510 --> 00:05:23,680 So when we see dots that are closer together, 70 00:05:23,680 --> 00:05:26,720 we interpret that to mean similar. 71 00:05:27,750 --> 00:05:30,200 So it's really important that we understand that, 72 00:05:30,200 --> 00:05:34,750 because you'll see that as you create visualization, it's very often 73 00:05:34,750 --> 00:05:40,730 easy to accidentally put things together that don't have the same semantic meaning. 74 00:05:40,730 --> 00:05:45,160 So we need to be very careful of this because when we see shapes, 75 00:05:45,160 --> 00:05:47,000 we infer meaning. 76 00:05:49,780 --> 00:05:54,690 Another example of proximity here, is the way that 77 00:05:54,690 --> 00:06:00,910 we just look at this distribution of shapes that are the human form. 78 00:06:00,910 --> 00:06:04,120 And because of the way that they're approximal to one another, 79 00:06:04,120 --> 00:06:09,140 we interpret this as movement and as a formation of people. 80 00:06:09,140 --> 00:06:12,580 But if you tried to describe this in computational terms, 81 00:06:12,580 --> 00:06:17,430 it's really just the same image, just in different orientations and 82 00:06:17,430 --> 00:06:19,320 locations relative to one another. 83 00:06:19,320 --> 00:06:22,546 But the human mind reads this as motion. 84 00:06:22,546 --> 00:06:23,819 A cloud. 85 00:06:25,595 --> 00:06:29,746 So let's talk about how the way that these features in 86 00:06:29,746 --> 00:06:35,660 the human perception affect the way that we should design visualizations. 87 00:06:37,310 --> 00:06:38,720 So I'll ask a question. 88 00:06:38,720 --> 00:06:42,630 Very simple, look at the following shapes and 89 00:06:42,630 --> 00:06:44,760 say which of the, which of them is bigger. 90 00:06:46,410 --> 00:06:51,360 So it's pretty clear that the there's a tall and short rectangle. 91 00:06:51,360 --> 00:06:52,970 Now here's this question. 92 00:06:52,970 --> 00:06:55,410 How much bigger is the tall rectangle? 93 00:06:56,590 --> 00:07:02,885 Well, the human mind can make this assessment relative much better, 94 00:07:02,885 --> 00:07:05,630 than it can this assessment. 95 00:07:07,000 --> 00:07:09,750 So interestingly, if we were to say, 96 00:07:09,750 --> 00:07:15,440 how much taller is the tall rectangle than the short rectangle, 97 00:07:16,630 --> 00:07:21,530 it's far easier for the human mind to see that actually it's three times as tall. 98 00:07:23,310 --> 00:07:25,330 Make the same determination of this circle. 99 00:07:26,770 --> 00:07:31,850 It's actually significantly harder for us to make that same determination. 100 00:07:31,850 --> 00:07:35,170 But interestingly, they are exactly the same heights. 101 00:07:36,865 --> 00:07:39,210 Let's talk about color for a moment. 102 00:07:40,968 --> 00:07:44,097 Let's ask the question, which of the following squares is brighter? 103 00:07:46,418 --> 00:07:49,090 Give everyone a minute, make a determination. 104 00:07:50,460 --> 00:07:51,370 Okay. 105 00:07:51,370 --> 00:07:56,950 So the answer is the one on the left is slightly brighter, but what 106 00:07:56,950 --> 00:08:01,600 you'll see is many people will actually not have been able to tell the difference. 107 00:08:01,600 --> 00:08:03,760 Let's look at another one. 108 00:08:03,760 --> 00:08:06,216 Look at this pair of images and say which is brighter. 109 00:08:08,341 --> 00:08:12,436 Actually, what you can see is the one on the right is brighter, 110 00:08:12,436 --> 00:08:15,090 if we describe them in RGB space. 111 00:08:15,090 --> 00:08:22,180 And this amount of difference is actually perceivable by a large number of people. 112 00:08:22,180 --> 00:08:26,440 So when we look at the way the human mind understands color, 113 00:08:27,510 --> 00:08:33,550 what we have to understand is that even though there are differences 114 00:08:33,550 --> 00:08:40,230 that we might be able to label in RGB space, the human mind 115 00:08:40,230 --> 00:08:46,910 is only capable of perceiving what we call the just noticeable difference. 116 00:08:46,910 --> 00:08:50,910 In other words, there're some steps in color space, 117 00:08:50,910 --> 00:08:54,690 that are not perceptible to the human mind. 118 00:08:54,690 --> 00:09:00,640 So what we have to realize, is that if we don't allow for 119 00:09:00,640 --> 00:09:07,540 enough space between individual color values in a scientific visualization, 120 00:09:07,540 --> 00:09:13,320 a human mind won't be able to perceive key differences in the data. 121 00:09:13,320 --> 00:09:15,190 This is a common, common, 122 00:09:15,190 --> 00:09:21,450 common problem you see in scientific visualization, that uses continuous color. 123 00:09:21,450 --> 00:09:25,580 It overlooks this very important fact of perception, 124 00:09:25,580 --> 00:09:31,950 which is that if you don't allow for humans to perceive the differences, 125 00:09:31,950 --> 00:09:38,010 your data will effectively become subsampled. 126 00:09:38,010 --> 00:09:42,237 Because people won't be able to see the difference between a value and 127 00:09:42,237 --> 00:09:45,388 another value that are close enough in color space, 128 00:09:45,388 --> 00:09:49,130 even though their differences are clear in numerical space. 129 00:09:50,741 --> 00:09:54,874 So let's think about, what is this mean, if we're going to 130 00:09:54,874 --> 00:10:00,240 use how the human mind perceives color, and bring this into a visualization. 131 00:10:02,944 --> 00:10:05,880 well, actually color and shape, et cetera. 132 00:10:05,880 --> 00:10:12,175 So, what we find important is this idea going back to McKinley. 133 00:10:12,175 --> 00:10:15,010 The one of the early visualization thinkers, 134 00:10:15,010 --> 00:10:17,140 is he put forth these heuristics. 135 00:10:17,140 --> 00:10:21,000 The first being something called the principal of consistency. 136 00:10:21,000 --> 00:10:23,890 And that means that the properties of the image 137 00:10:25,190 --> 00:10:27,400 should match the properties of the data. 138 00:10:28,530 --> 00:10:31,800 In other words if something's big, the data should make it look big. 139 00:10:31,800 --> 00:10:34,040 If it's small, the data should look, make it look small. 140 00:10:35,130 --> 00:10:40,820 And then the second idea is the importance of the principle of importance ordering. 141 00:10:40,820 --> 00:10:43,680 And this is something that we often overlook. 142 00:10:43,680 --> 00:10:47,830 The idea is that take the most important variables, and 143 00:10:47,830 --> 00:10:50,620 encode them in the most effective way. 144 00:10:50,620 --> 00:10:55,050 And our understanding of human perception, 145 00:10:56,330 --> 00:11:02,010 is such that we perceive differences of 146 00:11:04,870 --> 00:11:10,990 magnitude differently when they're encoded using different encoding methods. 147 00:11:10,990 --> 00:11:15,600 And so if you can see that position is much more 148 00:11:15,600 --> 00:11:19,760 clear of a difference to the human mind than color. 149 00:11:21,560 --> 00:11:26,490 So what this means is that when you are making a visualization, the shape, 150 00:11:26,490 --> 00:11:30,960 excuse me, the, the variables that are the most important, 151 00:11:30,960 --> 00:11:37,620 are often best to map to position and length rather than color, 152 00:11:37,620 --> 00:11:43,320 because it's hard for the human mind to perceive differences in color. 153 00:11:44,360 --> 00:11:48,930 And this is something that you see so very often in scientific visualization. 154 00:11:48,930 --> 00:11:53,560 That, someone will make an example where they're trying to show difference in 155 00:11:53,560 --> 00:11:59,140 volume, where the same numerical difference would be much 156 00:11:59,140 --> 00:12:05,350 more clearly demonstrated by differences in length, or in angle, or in slope. 157 00:12:06,850 --> 00:12:12,010 So you have to choose the dimensions that you have available, and 158 00:12:12,010 --> 00:12:17,860 think about the saliency that they have when people are perceiving what 159 00:12:17,860 --> 00:12:23,290 they're looking at, and think about it not you as the scientist, 160 00:12:23,290 --> 00:12:27,180 the creator of the visualization, but the human who is not you, 161 00:12:27,180 --> 00:12:31,520 who doesn't know the data, will look at something and 162 00:12:31,520 --> 00:12:35,330 not be able to perceive differences that are visible to you. 163 00:12:37,220 --> 00:12:40,090 So let's try and quantify this a little bit specifically. 164 00:12:42,550 --> 00:12:46,230 This'll be something that will probably have occurred in a number of 165 00:12:46,230 --> 00:12:48,430 the other lectures. 166 00:12:48,430 --> 00:12:53,520 If we take data and map them to different types. 167 00:12:55,100 --> 00:12:57,000 So the first is nominal. 168 00:12:57,000 --> 00:12:59,190 These are often category labels. 169 00:12:59,190 --> 00:13:02,040 So if you have data that describe fruits, 170 00:13:02,040 --> 00:13:05,622 the categories would be things like apples and oranges. 171 00:13:05,622 --> 00:13:10,624 There might be numerical data, but it's of an ordinal nature, or numerical or 172 00:13:10,624 --> 00:13:14,280 non-numerical data that's of an ordinal nature. 173 00:13:14,280 --> 00:13:17,080 So the example I put here is the quality of meat, 174 00:13:17,080 --> 00:13:21,140 where the example would be that you have A, and double A, and triple A. 175 00:13:21,140 --> 00:13:22,270 And what they mean is, 176 00:13:22,270 --> 00:13:26,820 that they are relative to one another, but magnitude is not important. 177 00:13:26,820 --> 00:13:29,130 Double A is probably not twice as good as A. 178 00:13:30,380 --> 00:13:32,230 And then lastly, qualitative, which, 179 00:13:32,230 --> 00:13:38,984 excuse me quantitative, which is numerical and often continuous values. 180 00:13:38,984 --> 00:13:40,529 That's something like a measure like length. 181 00:13:41,910 --> 00:13:47,025 So let's think about, if we have data that have certain types, 182 00:13:47,025 --> 00:13:53,460 that if we want to map those particular values to the way 183 00:13:53,460 --> 00:13:59,560 that the human mind perceives, you'll see that quantitative values, 184 00:13:59,560 --> 00:14:04,670 ordinal values and nominal values all map best to position. 185 00:14:06,210 --> 00:14:13,160 But you'll also see that if you, the next best value for nominal is hue. 186 00:14:14,650 --> 00:14:16,220 So that means color. 187 00:14:17,440 --> 00:14:23,350 So if you want to label categories, the, one of the best ways 188 00:14:23,350 --> 00:14:29,850 is differences in color, but I think we'll show in a moment, not continuous color. 189 00:14:29,850 --> 00:14:34,500 Whereas if you look at hue under quantitative, 190 00:14:34,500 --> 00:14:38,320 it's actually about two thirds of the way down the column. 191 00:14:39,390 --> 00:14:45,210 So if you want to show a number, and it has magnitude, and it's a quantitative 192 00:14:45,210 --> 00:14:50,350 number, and you're picking hue as the mapping, you should probably think again 193 00:14:50,350 --> 00:14:53,910 about that, because it's not going to be easy for the human mind to perceive. 194 00:14:56,470 --> 00:14:58,620 Let's pick a more specific example. 195 00:15:00,580 --> 00:15:01,990 color. 196 00:15:01,990 --> 00:15:05,940 So we've talked previously about this idea, that for, 197 00:15:05,940 --> 00:15:11,420 that hue is one of the important ways that we can encode a nominal value. 198 00:15:12,940 --> 00:15:17,610 But if you look here, what you can see is that the human mind, 199 00:15:17,610 --> 00:15:23,020 based on what we described earlier, and our ability to perceive differences, 200 00:15:23,020 --> 00:15:28,010 means that on a, a color scale like black to white, 201 00:15:28,010 --> 00:15:33,690 that the human mind is at best able to perceive seven differences. 202 00:15:35,230 --> 00:15:40,490 And so, if you look at the continuous values, you can see that co, hue can 203 00:15:40,490 --> 00:15:46,660 encode continuous values, but it's far less good than its ability 204 00:15:46,660 --> 00:15:53,571 to include ordinal values, which you can use with different magnitude, 205 00:15:53,571 --> 00:15:57,960 with hue of a single monotonically changing function. 206 00:15:59,930 --> 00:16:01,920 Another thing that I think you can use, 207 00:16:01,920 --> 00:16:06,120 is that hue is normally perceived as unordered. 208 00:16:07,440 --> 00:16:15,390 And so, this means that hue is a really good way to label nominal variables. 209 00:16:15,390 --> 00:16:20,840 And so, when you have things that are, have no ordering, picking color is 210 00:16:20,840 --> 00:16:26,065 really valuable, because color, hue does not have an order. 211 00:16:26,065 --> 00:16:31,680 Whereas if you look at color value, it does have an order. 212 00:16:31,680 --> 00:16:36,640 And so, it effectively is not nearly as good for 213 00:16:36,640 --> 00:16:40,690 nominal values, than the unordered color values. 214 00:16:42,380 --> 00:16:45,380 And we could show how these are applied if we 215 00:16:45,380 --> 00:16:49,330 look at different kinds of the way that color is used. 216 00:16:49,330 --> 00:16:54,940 So if we look at for example in this illustration from an anatomy textbook, 217 00:16:54,940 --> 00:17:00,670 how different anatomical features are colored using different values. 218 00:17:00,670 --> 00:17:03,730 So here, red might mean artery and 219 00:17:03,730 --> 00:17:08,590 blue might mean vein and yellow might mean capillary. 220 00:17:08,590 --> 00:17:11,520 But they're effectively nominal values. 221 00:17:11,520 --> 00:17:13,339 So colors being used nominally. 222 00:17:14,640 --> 00:17:19,070 You can look similarly at this presidential election map from the from 223 00:17:19,070 --> 00:17:22,430 NBC, and you can see here that we've also used color in 224 00:17:22,430 --> 00:17:26,950 a way where red is mapped to the Republicans and blue to the Democrats. 225 00:17:26,950 --> 00:17:30,490 So here, we've just simply taken two colors and 226 00:17:30,490 --> 00:17:32,389 made a very easily distinguishable map. 227 00:17:33,570 --> 00:17:39,200 But if we look at quantitative uses of color, and this is where the most 228 00:17:40,650 --> 00:17:44,990 often you most often see mistakes in Earth 229 00:17:44,990 --> 00:17:50,050 science data visualizations by quantitative applications of color. 230 00:17:51,070 --> 00:17:57,760 So here, what you have is an example of Global Average Temperature and 231 00:17:57,760 --> 00:18:01,578 it's rendered from blue to red. 232 00:18:01,578 --> 00:18:05,970 Now there're a number of reasons based on what we just learned about color, 233 00:18:05,970 --> 00:18:09,930 that described why this is not the best mapping. 234 00:18:09,930 --> 00:18:13,670 So first of all, the, there's a, it's very hard for 235 00:18:13,670 --> 00:18:19,610 the human mind to perceive more than seven seven different colors. 236 00:18:19,610 --> 00:18:24,190 So it's really hard for a hum, for a person to look along the equator and 237 00:18:24,190 --> 00:18:29,920 tell me the different average temperature between Panama and Senegal. 238 00:18:29,920 --> 00:18:35,610 And they're probably not the same, but in red, they're relatively close. 239 00:18:35,610 --> 00:18:39,060 So it's hard for the eye to distinguish values like that. 240 00:18:39,060 --> 00:18:42,780 Another example here is this fluid dynamics simulation, 241 00:18:42,780 --> 00:18:45,110 where velocity is encoded as color. 242 00:18:46,300 --> 00:18:50,530 And what you see here is that these hues, there is no natural ordering. 243 00:18:50,530 --> 00:18:53,230 If you look at the yellow in the middle, 244 00:18:53,230 --> 00:18:57,371 it's really hard to tell where that lies in color space. 245 00:18:57,371 --> 00:19:02,360 And different, our ability to perceive light 246 00:19:02,360 --> 00:19:06,689 is emphasizes different kinds of scalar values. 247 00:19:08,310 --> 00:19:12,750 So if you look at the way that color can be applied in an ordinal fashion, 248 00:19:14,050 --> 00:19:21,160 it's very useful to pick a a single saturation, 249 00:19:21,160 --> 00:19:23,220 excuse me, to vary luminance and 250 00:19:23,220 --> 00:19:29,370 saturation together, and you get these very nicely marked color scales. 251 00:19:29,370 --> 00:19:33,390 Where what you're able to see very clearly are the demarcations between each of 252 00:19:33,390 --> 00:19:34,370 the different values. 253 00:19:35,790 --> 00:19:39,130 And you can do the same for different types of color, 254 00:19:39,130 --> 00:19:43,570 where if color is diverging, in other words, if you want to emphasize the mid 255 00:19:43,570 --> 00:19:48,520 point of a color range, you can pick a neutral color in the middle. 256 00:19:48,520 --> 00:19:50,740 And when you apply this to a map, 257 00:19:50,740 --> 00:19:55,120 you'll see that the midpoint that use saturated colors on 258 00:19:55,120 --> 00:19:59,160 the different end points, that they'll be distinct from the midpoint of the map. 259 00:20:00,550 --> 00:20:05,760 So just at a high level review, if you're going to be thinking about 260 00:20:05,760 --> 00:20:10,830 color in data visualization, you should use only a few colors and 261 00:20:10,830 --> 00:20:15,450 that, try to be able to make them as distinct as possible. 262 00:20:15,450 --> 00:20:19,820 Strive for harmony, be very cautious of cultural conventions. 263 00:20:21,290 --> 00:20:24,750 Be cautious about bad interactions between colors. 264 00:20:24,750 --> 00:20:28,462 And one of the best ways to determine if your visualization is going to be 265 00:20:28,462 --> 00:20:30,510 effective, is before adding color, 266 00:20:30,510 --> 00:20:34,929 see if without it you can distinguish the different variables in black and white. 267 00:20:36,133 --> 00:20:37,180 We'll ahead into the next chapter.