WEBVTT

1
00:00:03.465 --> 00:00:06.845
<00:00:03.465>So <00:00:03.639>welcome <00:00:03.815>everybody <00:00:03.989>to <00:00:04.165>Machine <00:00:04.340>Learning <00:00:04.515>1 <00:00:04.689>0 <00:00:04.865>1.

2
00:00:07.135 --> 00:00:12.045
<00:00:07.135>Where <00:00:07.368>to <00:00:07.601>begin? <00:00:07.835>Um, <00:00:08.068>you're <00:00:08.301>probably <00:00:08.535>here,

3
00:00:12.395 --> 00:00:14.085
<00:00:12.395>well <00:00:12.861>going <00:00:13.328>very <00:00:13.795>fast.

4
00:00:14.705 --> 00:00:19.045
<00:00:14.705>No, <00:00:14.938>sorry <00:00:15.171>for <00:00:15.405>that. <00:00:15.638>Just <00:00:15.871>start <00:00:16.105>again.

5
00:00:22.915 --> 00:00:24.845
<00:00:22.915>Technic, <00:00:23.615>right? <00:00:24.314>Yes.

6
00:00:26.185 --> 00:00:28.445
<00:00:26.185>So <00:00:26.384>you <00:00:26.584>probably <00:00:26.785>are <00:00:26.985>all <00:00:27.185>here <00:00:27.384>with <00:00:27.584>ideas.

7
00:00:28.765 --> 00:00:32.965
<00:00:28.765>I <00:00:28.920>wanna <00:00:29.076>do <00:00:29.231>something <00:00:29.387>with <00:00:29.542>AI <00:00:29.698>or <00:00:29.853>I'm <00:00:30.009>machine <00:00:30.165>learning,

8
00:00:33.025 --> 00:00:35.005
<00:00:33.025>but <00:00:33.225>I <00:00:33.425>have <00:00:33.625>no <00:00:33.824>clue <00:00:34.025>where <00:00:34.225>to <00:00:34.425>start.

9
00:00:36.065 --> 00:00:39.005
<00:00:36.065>So <00:00:36.415>that's <00:00:36.765>why <00:00:37.114>you're <00:00:37.464>here.

10
00:00:39.025 --> 00:00:41.125
<00:00:39.025>And <00:00:39.199>if <00:00:39.375>you're <00:00:39.550>already <00:00:39.725>an <00:00:39.900>expert <00:00:40.074>on <00:00:40.250>machine <00:00:40.425>learning,

11
00:00:41.125 --> 00:00:43.645
<00:00:41.125>welcome, <00:00:41.280>but <00:00:41.436>I <00:00:41.591>don't <00:00:41.747>think <00:00:41.902>you're <00:00:42.058>at <00:00:42.213>the <00:00:42.369>right <00:00:42.525>place.

12
00:00:45.285 --> 00:00:47.765
<00:00:45.285>I <00:00:45.459>had <00:00:45.635>the <00:00:45.809>same <00:00:45.985>idea <00:00:46.160>like <00:00:46.334>seven <00:00:46.510>years <00:00:46.684>ago.

13
00:00:48.685 --> 00:00:50.805
<00:00:48.685>I <00:00:48.918>had <00:00:49.151>two <00:00:49.385>cats, <00:00:49.618>Phineas <00:00:49.851>and <00:00:50.085>her,

14
00:00:51.425 --> 00:00:53.725
<00:00:51.425>and <00:00:51.705>they <00:00:51.985>had <00:00:52.265>different <00:00:52.544>cat <00:00:52.824>foods.

15
00:00:54.625 --> 00:00:56.885
<00:00:54.625>So <00:00:54.858>the <00:00:55.091>one <00:00:55.325>was <00:00:55.558>the <00:00:55.791>kitten <00:00:56.025>foods,

16
00:00:57.105 --> 00:00:59.845
<00:00:57.105>but <00:00:57.260>a <00:00:57.416>big <00:00:57.571>one <00:00:57.727>also <00:00:57.882>want <00:00:58.038>to <00:00:58.193>eat <00:00:58.349>big, <00:00:58.504>uh,

17
00:00:59.845 --> 00:01:01.565
<00:00:59.845>the <00:01:00.078>kitten <00:01:00.311>foods <00:01:00.545>but <00:01:00.778>he <00:01:01.011>grows <00:01:01.245>back.

18
00:01:02.025 --> 00:01:04.285
<00:01:02.025>So <00:01:02.199>I <00:01:02.375>thought <00:01:02.550>I <00:01:02.725>need <00:01:02.900>a <00:01:03.074>solution <00:01:03.250>for <00:01:03.425>this.

19
00:01:05.105 --> 00:01:09.565
<00:01:05.105>So <00:01:05.260>I <00:01:05.416>went <00:01:05.571>to <00:01:05.727>the <00:01:05.882>Ikea, <00:01:06.038>bought <00:01:06.193>uh, <00:01:06.349>a <00:01:06.505>thing,

20
00:01:10.225 --> 00:01:11.805
<00:01:10.225>put <00:01:10.399>some <00:01:10.574>hole <00:01:10.750>in <00:01:10.925>it, <00:01:11.100>and <00:01:11.274>a <00:01:11.449>cat <00:01:11.625>flap

21
00:01:12.145 --> 00:01:16.765
<00:01:12.145>and <00:01:12.261>I <00:01:12.378>can <00:01:12.494>think, <00:01:12.611>yeah, <00:01:12.728>I <00:01:12.845>can <00:01:12.961>read <00:01:13.078>the <00:01:13.195>chip <00:01:13.311>in <00:01:13.428>their <00:01:13.545>neck

22
00:01:17.185 --> 00:01:18.365
<00:01:17.185>and <00:01:17.397>see <00:01:17.609>which <00:01:17.822>cat <00:01:18.034>it <00:01:18.247>is.

23
00:01:18.785 --> 00:01:20.885
<00:01:18.785>But <00:01:18.960>I <00:01:19.134>wanted <00:01:19.310>to <00:01:19.485>do <00:01:19.660>something <00:01:19.835>with <00:01:20.009>machine <00:01:20.185>learning.

24
00:01:21.585 --> 00:01:26.205
<00:01:21.585>So <00:01:21.935>let's <00:01:22.285>over-engineer <00:01:22.635>this. <00:01:22.985>Yes.

25
00:01:27.185 --> 00:01:30.805
<00:01:27.185>So <00:01:27.301>I <00:01:27.418>had <00:01:27.535>a <00:01:27.651>lot <00:01:27.768>of <00:01:27.885>pictures <00:01:28.001>of <00:01:28.118>cats <00:01:28.235>already <00:01:28.351>on <00:01:28.468>my <00:01:28.585>phone

26
00:01:31.665 --> 00:01:34.405
<00:01:31.664>and <00:01:31.839>I <00:01:32.014>trained <00:01:32.190>a <00:01:32.365>model <00:01:32.539>to <00:01:32.714>recognize <00:01:32.889>the <00:01:33.065>cats

27
00:01:34.545 --> 00:01:36.925
<00:01:34.545>to <00:01:34.720>give <00:01:34.895>them <00:01:35.070>access <00:01:35.245>to <00:01:35.420>the <00:01:35.595>correct <00:01:35.770>bowl <00:01:35.945>foods.

28
00:01:38.625 --> 00:01:40.645
<00:01:38.625>Uh, <00:01:38.858>we're <00:01:39.091>gonna <00:01:39.325>talk <00:01:39.558>a <00:01:39.791>lot <00:01:40.025>today

29
00:01:40.705 --> 00:01:43.645
<00:01:40.705>and <00:01:40.832>I <00:01:40.959>only <00:01:41.086>have <00:01:41.214>30 <00:01:41.341>minutes, <00:01:41.468>I'm <00:01:41.595>just <00:01:41.723>going <00:01:41.850>through <00:01:41.977>with <00:01:42.105>it.

30
00:01:45.025 --> 00:01:47.485
<00:01:45.025>Uh, <00:01:45.258>I <00:01:45.491>am, <00:01:45.725>I <00:01:45.958>work <00:01:46.191>for <00:01:46.425>Luminous.

31
00:01:48.025 --> 00:01:51.045
<00:01:48.025>I'm <00:01:48.200>almost <00:01:48.375>nine <00:01:48.550>years <00:01:48.725>in, <00:01:48.900>uh, <00:01:49.075>in <00:01:49.250>the <00:01:49.425>business.

32
00:01:51.795 --> 00:01:54.125
<00:01:51.795>Most <00:01:51.970>of <00:01:52.145>the <00:01:52.320>time <00:01:52.495>I <00:01:52.670>do <00:01:52.845>jaff <00:01:53.020>and <00:01:53.195>angler,

33
00:01:54.305 --> 00:01:56.485
<00:01:54.305>but <00:01:54.505>uh, <00:01:54.705>yeah, <00:01:54.905>machine <00:01:55.105>learning <00:01:55.305>is <00:01:55.505>like <00:01:55.705>a

34
00:01:56.485 --> 00:01:57.605
<00:01:56.485>passion <00:01:56.821>project <00:01:57.157>for <00:01:57.493>me.

35
00:01:58.185 --> 00:02:00.165
<00:01:58.185>So <00:01:58.385>that's <00:01:58.585>why <00:01:58.785>I'm <00:01:58.985>talking <00:01:59.185>about <00:01:59.385>C <00:01:59.585>Learning

36
00:02:01.155 --> 00:02:02.245
<00:02:01.155>yuca <00:02:01.482>scan <00:02:01.809>the <00:02:02.136>codes

37
00:02:02.705 --> 00:02:05.325
<00:02:02.705>and <00:02:02.860>I <00:02:03.016>work <00:02:03.171>for <00:02:03.327>Luminous, <00:02:03.482>that's <00:02:03.638>a, <00:02:03.793>that <00:02:03.949>consultancy <00:02:04.105>company.

38
00:02:06.785 --> 00:02:10.285
<00:02:06.785>So <00:02:07.134>what <00:02:07.485>is <00:02:07.835>machine <00:02:08.185>learning?

39
00:02:11.515 --> 00:02:14.725
<00:02:11.515>Machine <00:02:11.714>learning <00:02:11.915>is <00:02:12.114>basically <00:02:12.315>teaching <00:02:12.515>the <00:02:12.714>the <00:02:12.915>computer

40
00:02:14.725 --> 00:02:16.965
<00:02:14.725>something <00:02:15.191>without <00:02:15.658>programming <00:02:16.125>it.

41
00:02:18.465 --> 00:02:20.885
<00:02:18.465>And <00:02:18.620>then <00:02:18.776>we <00:02:18.931>have <00:02:19.087>AI <00:02:19.242>and <00:02:19.398>it's <00:02:19.553>all <00:02:19.709>the <00:02:19.865>conference.

42
00:02:20.945 --> 00:02:23.085
<00:02:20.945>So <00:02:21.144>I <00:02:21.345>would <00:02:21.545>skip <00:02:21.745>gut <00:02:21.945>right <00:02:22.144>through <00:02:22.345>it.

43
00:02:23.425 --> 00:02:25.965
<00:02:23.425>That's <00:02:23.658>more <00:02:23.891>about <00:02:24.125>imitating <00:02:24.358>the <00:02:24.591>human <00:02:24.825>brain.

44
00:02:27.045 --> 00:02:29.045
<00:02:27.045>MC <00:02:27.278>learning <00:02:27.511>is <00:02:27.745>a <00:02:27.978>subset <00:02:28.211>of <00:02:28.445>that.

45
00:02:30.225 --> 00:02:31.805
<00:02:30.225>And <00:02:30.505>then <00:02:30.785>we <00:02:31.065>have <00:02:31.345>deep <00:02:31.625>learning.

46
00:02:31.905 --> 00:02:34.485
<00:02:31.905>And <00:02:32.105>deep <00:02:32.305>learning <00:02:32.505>is <00:02:32.705>all <00:02:32.905>about <00:02:33.105>neural <00:02:33.305>networks.

47
00:02:35.525 --> 00:02:38.685
<00:02:35.525>Everyone <00:02:35.680>has <00:02:35.836>a <00:02:35.991>neural <00:02:36.147>network <00:02:36.302>in <00:02:36.458>his <00:02:36.613>or <00:02:36.769>her <00:02:36.925>brain.

48
00:02:40.025 --> 00:02:43.885
<00:02:40.025>And <00:02:40.180>a <00:02:40.336>neural <00:02:40.491>network <00:02:40.647>consists <00:02:40.802>of <00:02:40.958>different <00:02:41.113>kinds <00:02:41.269>of <00:02:41.425>neurons

49
00:02:43.905 --> 00:02:47.645
<00:02:43.905>and <00:02:44.060>every <00:02:44.216>neuron <00:02:44.371>is <00:02:44.527>very <00:02:44.682>good <00:02:44.838>at <00:02:44.993>one <00:02:45.149>specific <00:02:45.305>task.

50
00:02:48.865 --> 00:02:50.565
<00:02:48.865>And <00:02:49.098>that <00:02:49.331>combined <00:02:49.565>is <00:02:49.798>the <00:02:50.031>neural <00:02:50.265>network.

51
00:02:50.825 --> 00:02:51.965
<00:02:50.825>But <00:02:51.030>that's <00:02:51.235>a <00:02:51.440>little <00:02:51.645>bit <00:02:51.851>deeper.

52
00:02:52.225 --> 00:02:53.605
<00:02:52.225>Not <00:02:52.473>gonna <00:02:52.721>talk <00:02:52.970>about <00:02:53.218>that <00:02:53.466>today.

53
00:02:54.845 --> 00:02:59.445
<00:02:54.845>I <00:02:55.000>have <00:02:55.156>gen <00:02:55.311>ai, <00:02:55.467>generative <00:02:55.622>AI <00:02:55.778>about <00:02:55.933>making <00:02:56.089>the <00:02:56.245>pictures

54
00:02:59.585 --> 00:03:01.925
<00:02:59.585>and <00:02:59.935>stuff <00:03:00.285>and <00:03:00.635>art <00:03:00.985>generated,

55
00:03:02.905 --> 00:03:05.045
<00:03:02.905>but <00:03:03.105>also <00:03:03.305>lots <00:03:03.505>of <00:03:03.705>talk <00:03:03.905>about <00:03:04.105>that <00:03:04.305>today.

56
00:03:06.065 --> 00:03:09.165
<00:03:06.065>And <00:03:06.204>if <00:03:06.345>you <00:03:06.484>see <00:03:06.625>them <00:03:06.765>overall, <00:03:06.905>lemme <00:03:07.045>see. <00:03:07.185>AI <00:03:07.325>is <00:03:07.465>everything.

57
00:03:09.315 --> 00:03:12.725
<00:03:09.315>Then <00:03:09.490>we <00:03:09.665>have <00:03:09.840>machine <00:03:10.015>learning, <00:03:10.190>deep <00:03:10.365>learning <00:03:10.540>and <00:03:10.715>ai.

58
00:03:14.545 --> 00:03:16.965
<00:03:14.545>And <00:03:14.744>you <00:03:14.945>probably <00:03:15.144>also <00:03:15.345>heard <00:03:15.545>about <00:03:15.744>data <00:03:15.945>science

59
00:03:17.185 --> 00:03:20.125
<00:03:17.185>and <00:03:17.418>it's <00:03:17.651>more <00:03:17.885>about <00:03:18.118>extracting <00:03:18.351>the <00:03:18.585>knowledge

60
00:03:20.625 --> 00:03:22.605
<00:03:20.625>and <00:03:20.905>user <00:03:21.185>to <00:03:21.465>predict <00:03:21.745>and <00:03:22.025>advice.

61
00:03:23.905 --> 00:03:25.325
<00:03:23.905>And <00:03:24.118>that's <00:03:24.331>more <00:03:24.543>to <00:03:24.757>the <00:03:24.970>side <00:03:25.183>effects.

62
00:03:27.445 --> 00:03:31.105
<00:03:27.445>So <00:03:27.644>we <00:03:27.845>have <00:03:28.045>different <00:03:28.245>types <00:03:28.445>of <00:03:28.644>machine <00:03:28.845>learning.

63
00:03:32.905 --> 00:03:35.845
<00:03:32.905>We <00:03:33.185>have <00:03:33.465>the <00:03:33.745>supervised, <00:03:34.025>the <00:03:34.305>unsupervised,

64
00:03:35.845 --> 00:03:38.965
<00:03:35.845>the <00:03:36.125>semi-supervised <00:03:36.405>and <00:03:36.685>the <00:03:36.965>reinforcement <00:03:37.245>learning.

65
00:03:40.305 --> 00:03:44.005
<00:03:40.305>So <00:03:40.538>with <00:03:40.771>supervised <00:03:41.005>learning <00:03:41.238>you <00:03:41.471>have <00:03:41.705>pictures,

66
00:03:44.025 --> 00:03:46.445
<00:03:44.025>for <00:03:44.165>example <00:03:44.305>a <00:03:44.445>cat <00:03:44.585>and <00:03:44.725>you <00:03:44.865>label <00:03:45.005>it <00:03:45.145>as <00:03:45.285>a <00:03:45.425>cat

67
00:03:47.505 --> 00:03:50.565
<00:03:47.505>and <00:03:47.621>you <00:03:47.738>have <00:03:47.855>a <00:03:47.971>duck <00:03:48.088>and <00:03:48.204>you <00:03:48.321>label <00:03:48.438>it, <00:03:48.555>it's <00:03:48.671>not <00:03:48.788>a <00:03:48.905>duck.

68
00:03:50.915 --> 00:03:53.485
<00:03:50.915>Then <00:03:51.054>you <00:03:51.195>train <00:03:51.334>a <00:03:51.475>model <00:03:51.614>and <00:03:51.755>you <00:03:51.894>give <00:03:52.035>it <00:03:52.174>a <00:03:52.315>picture

69
00:03:53.745 --> 00:03:56.685
<00:03:53.745>and <00:03:53.920>it <00:03:54.095>predicts <00:03:54.270>it's <00:03:54.445>a <00:03:54.620>cat <00:03:54.795>or <00:03:54.970>something <00:03:55.145>else.

70
00:03:57.625 --> 00:03:59.805
<00:03:57.625>As <00:03:57.800>you <00:03:57.975>see <00:03:58.150>by <00:03:58.325>now, <00:03:58.500>I <00:03:58.675>really <00:03:58.850>love <00:03:59.025>cats.

71
00:04:00.385 --> 00:04:02.925
<00:04:00.385>If <00:04:00.584>you <00:04:00.785>don't <00:04:00.984>love <00:04:01.185>cats, <00:04:01.385>just <00:04:01.584>predict, <00:04:01.785>uh,

72
00:04:03.315 --> 00:04:04.845
<00:04:03.315>just <00:04:03.659>pretend <00:04:04.003>it's <00:04:04.347>a <00:04:04.692>duck.

73
00:04:07.575 --> 00:04:09.375
<00:04:07.575>Unsupervised <00:04:07.924>learning <00:04:08.274>is <00:04:08.625>the <00:04:08.975>opposite.

74
00:04:09.435 --> 00:04:11.855
<00:04:09.435>So <00:04:09.575>we <00:04:09.715>just <00:04:09.855>giving <00:04:09.995>it <00:04:10.135>a <00:04:10.275>bench <00:04:10.415>through <00:04:10.555>data, <00:04:10.695>good <00:04:10.835>luck

75
00:04:11.855 --> 00:04:13.575
<00:04:11.855>with <00:04:12.204>it, <00:04:12.554>train <00:04:12.905>the <00:04:13.255>model

76
00:04:13.995 --> 00:04:17.455
<00:04:13.995>and <00:04:14.170>then <00:04:14.345>it's <00:04:14.520>gonna <00:04:14.695>do <00:04:14.870>some <00:04:15.045>categorizing <00:04:15.220>for <00:04:15.395>you.

77
00:04:18.275 --> 00:04:19.775
<00:04:18.274>So <00:04:18.612>that's <00:04:18.950>also <00:04:19.287>an <00:04:19.625>option.

78
00:04:21.235 --> 00:04:23.215
<00:04:21.235>Semi-supervised <00:04:21.585>learning <00:04:21.935>is <00:04:22.285>more <00:04:22.635>the

79
00:04:23.215 --> 00:04:24.375
<00:04:23.215>combination <00:04:23.563>of <00:04:23.911>those <00:04:24.259>two.

80
00:04:24.395 --> 00:04:26.215
<00:04:24.395>So <00:04:24.674>you <00:04:24.955>have <00:04:25.234>label <00:04:25.515>data <00:04:25.794>af

81
00:04:26.235 --> 00:04:28.815
<00:04:26.235>and <00:04:26.435>label <00:04:26.635>data <00:04:26.835>give <00:04:27.035>to <00:04:27.235>the <00:04:27.435>model <00:04:27.635>training

82
00:04:29.035 --> 00:04:30.295
<00:04:29.035>and <00:04:29.318>it <00:04:29.602>gives <00:04:29.885>you <00:04:30.169>prediction.

83
00:04:33.535 --> 00:04:37.095
<00:04:33.534>Reinforcement <00:04:33.768>learning <00:04:34.001>is <00:04:34.234>the <00:04:34.468>next <00:04:34.701>one <00:04:34.934>there.

84
00:04:37.115 --> 00:04:38.815
<00:04:37.115>We, <00:04:37.395>it's <00:04:37.675>all <00:04:37.955>about <00:04:38.235>the <00:04:38.515>environments.

85
00:04:38.995 --> 00:04:41.375
<00:04:38.995>So <00:04:39.150>we <00:04:39.306>have <00:04:39.461>a <00:04:39.617>cat <00:04:39.772>and <00:04:39.928>it <00:04:40.083>has <00:04:40.239>two <00:04:40.395>options.

86
00:04:42.195 --> 00:04:44.615
<00:04:42.195>Doo <00:04:42.395>gonna <00:04:42.594>work <00:04:42.795>or <00:04:42.995>I'm <00:04:43.195>going <00:04:43.395>to <00:04:43.594>sit.

87
00:04:45.315 --> 00:04:48.335
<00:04:45.315>So <00:04:45.455>it <00:04:45.594>chooses <00:04:45.735>one <00:04:45.875>from <00:04:46.015>this, <00:04:46.155>that <00:04:46.295>case <00:04:46.435>it's <00:04:46.575>gonna <00:04:46.715>sit

88
00:04:48.995 --> 00:04:50.615
<00:04:48.995>and <00:04:49.195>then <00:04:49.395>it's <00:04:49.595>get <00:04:49.795>rewarded <00:04:49.995>with <00:04:50.195>a <00:04:50.395>fish

89
00:04:51.075 --> 00:04:53.575
<00:04:51.075>and <00:04:51.354>thinks, <00:04:51.635>hey, <00:04:51.914>I <00:04:52.195>like <00:04:52.474>fish.

90
00:04:54.075 --> 00:04:55.655
<00:04:54.075>So <00:04:54.425>it <00:04:54.775>updated <00:04:55.125>his <00:04:55.474>policy

91
00:04:56.075 --> 00:04:58.815
<00:04:56.075>and <00:04:56.215>the <00:04:56.354>next <00:04:56.495>time <00:04:56.635>it's <00:04:56.775>more <00:04:56.914>likely <00:04:57.055>it's <00:04:57.195>going <00:04:57.335>to <00:04:57.474>sit

92
00:04:59.355 --> 00:05:00.935
<00:04:59.355>and <00:04:59.588>then <00:04:59.821>it's <00:05:00.055>doing <00:05:00.288>again <00:05:00.521>and <00:05:00.755>again.

93
00:05:00.995 --> 00:05:03.615
<00:05:00.995>So <00:05:01.170>it's <00:05:01.345>really <00:05:01.520>useful <00:05:01.695>if <00:05:01.870>you <00:05:02.045>have <00:05:02.220>an <00:05:02.395>environment

94
00:05:03.615 --> 00:05:05.055
<00:05:03.615>that's <00:05:04.263>constantly <00:05:04.911>changing.

95
00:05:08.525 --> 00:05:10.575
<00:05:08.525>Then <00:05:08.758>we <00:05:08.991>have <00:05:09.224>different <00:05:09.458>types <00:05:09.691>of <00:05:09.924>data.

96
00:05:10.755 --> 00:05:13.735
<00:05:10.755>We <00:05:10.988>have <00:05:11.221>numerical, <00:05:11.455>categorial <00:05:11.688>and <00:05:11.921>text <00:05:12.155>data.

97
00:05:14.655 --> 00:05:17.735
<00:05:14.655>Numerical <00:05:14.888>is <00:05:15.121>more <00:05:15.354>about <00:05:15.588>different <00:05:15.821>data <00:05:16.054>points

98
00:05:18.835 --> 00:05:21.375
<00:05:18.835>and <00:05:19.010>uh, <00:05:19.185>voices <00:05:19.359>and <00:05:19.534>all <00:05:19.710>that <00:05:19.885>kind <00:05:20.060>of <00:05:20.234>things.

99
00:05:21.915 --> 00:05:25.335
<00:05:21.915>The <00:05:22.195>categorial <00:05:22.475>predefined <00:05:22.755>categories <00:05:23.035>like <00:05:23.315>uh,

100
00:05:25.735 --> 00:05:27.575
<00:05:25.735>countries <00:05:26.085>and <00:05:26.435>the <00:05:26.785>text <00:05:27.135>data.

101
00:05:27.635 --> 00:05:31.015
<00:05:27.635>And <00:05:27.775>that's <00:05:27.914>more <00:05:28.055>like <00:05:28.195>lots <00:05:28.335>of <00:05:28.474>text <00:05:28.615>in <00:05:28.755>files <00:05:28.895>and <00:05:29.034>stuff.

102
00:05:33.715 --> 00:05:37.135
<00:05:33.715>But <00:05:33.915>before <00:05:34.115>we <00:05:34.315>begin, <00:05:34.515>just <00:05:34.715>like <00:05:34.915>every <00:05:35.115>normal

103
00:05:37.855 --> 00:05:39.895
<00:05:37.855>software <00:05:38.055>problem, <00:05:38.255>we <00:05:38.455>have <00:05:38.655>to <00:05:38.855>understand <00:05:39.055>the <00:05:39.255>problem.

104
00:05:41.645 --> 00:05:43.185
<00:05:41.645>So <00:05:41.922>what's <00:05:42.199>the <00:05:42.476>nature <00:05:42.753>of <00:05:43.031>it?

105
00:05:43.185 --> 00:05:44.385
<00:05:43.185>What <00:05:43.365>do <00:05:43.545>we <00:05:43.725>wanna <00:05:43.905>achieve <00:05:44.085>with <00:05:44.265>it

106
00:05:45.405 --> 00:05:47.265
<00:05:45.405>and <00:05:45.684>what <00:05:45.965>do <00:05:46.244>we <00:05:46.525>have <00:05:46.804>available?

107
00:05:48.335 --> 00:05:51.465
<00:05:48.335>Because <00:05:48.490>most <00:05:48.646>of <00:05:48.801>the <00:05:48.957>times <00:05:49.112>when <00:05:49.268>customers <00:05:49.423>come <00:05:49.579>to <00:05:49.734>me

108
00:05:51.465 --> 00:05:54.705
<00:05:51.465>and <00:05:51.605>they <00:05:51.745>ask, <00:05:51.885>yeah, <00:05:52.025>I <00:05:52.165>wanna <00:05:52.305>do <00:05:52.445>something <00:05:52.585>with <00:05:52.725>uh, <00:05:52.865>AI

109
00:05:54.705 --> 00:05:55.905
<00:05:54.705>and <00:05:54.921>steel <00:05:55.137>learning, <00:05:55.352>can <00:05:55.568>you <00:05:55.784>help?

110
00:05:55.965 --> 00:05:58.005
<00:05:55.965>And <00:05:56.140>I'm <00:05:56.315>like, <00:05:56.490>yeah, <00:05:56.665>sure, <00:05:56.840>what <00:05:57.015>do <00:05:57.190>you <00:05:57.365>want?

111
00:05:58.225 --> 00:05:59.365
<00:05:58.225>Yes, <00:05:58.567>something <00:05:58.909>with <00:05:59.251>AI

112
00:05:59.505 --> 00:06:03.485
<00:05:59.505>and <00:05:59.660>machine <00:05:59.816>learning, <00:05:59.971>I'm <00:06:00.127>like, <00:06:00.282>yes, <00:06:00.438>of <00:06:00.593>course, <00:06:00.749>but <00:06:00.905>why?

113
00:06:03.625 --> 00:06:05.925
<00:06:03.625>What's <00:06:03.825>your <00:06:04.025>problem? <00:06:04.225>What <00:06:04.425>do <00:06:04.625>you <00:06:04.825>wanna <00:06:05.025>solve?

114
00:06:06.465 --> 00:06:08.845
<00:06:06.465>And <00:06:06.640>then <00:06:06.815>you <00:06:06.989>ask <00:06:07.164>the <00:06:07.340>typically <00:06:07.515>10 <00:06:07.690>times <00:06:07.864>why?

115
00:06:09.105 --> 00:06:11.365
<00:06:09.105>And <00:06:09.305>then <00:06:09.505>it's <00:06:09.705>like, <00:06:09.905>yeah, <00:06:10.105>we <00:06:10.305>only <00:06:10.505>want

116
00:06:11.365 --> 00:06:12.405
<00:06:11.365>to <00:06:11.521>use <00:06:11.677>it <00:06:11.832>because <00:06:11.989>of <00:06:12.145>the <00:06:12.301>hype.

117
00:06:12.405 --> 00:06:14.245
<00:06:12.405>And <00:06:12.604>I'm <00:06:12.804>like, <00:06:13.005>okay, <00:06:13.205>let's <00:06:13.405>do <00:06:13.604>it <00:06:13.804>cheaper

118
00:06:14.245 --> 00:06:17.085
<00:06:14.245>and <00:06:14.525>better <00:06:14.805>solution <00:06:15.085>without <00:06:15.365>it <00:06:15.645>sadly.

119
00:06:17.585 --> 00:06:18.725
<00:06:17.585>But <00:06:17.927>that's <00:06:18.269>the <00:06:18.611>case.

120
00:06:20.585 --> 00:06:22.645
<00:06:20.585>So <00:06:21.051>in <00:06:21.518>my <00:06:21.984>case,

121
00:06:23.225 --> 00:06:25.925
<00:06:23.225>the <00:06:23.425>problem <00:06:23.625>is <00:06:23.825>I <00:06:24.025>have <00:06:24.225>two <00:06:24.425>different <00:06:24.625>cats

122
00:06:25.995 --> 00:06:28.525
<00:06:25.995>with <00:06:26.195>different <00:06:26.395>foods <00:06:26.595>and <00:06:26.795>eating <00:06:26.995>each <00:06:27.195>other <00:06:27.395>foods.

123
00:06:29.605 --> 00:06:32.605
<00:06:29.605>I <00:06:29.760>have <00:06:29.916>a <00:06:30.071>lots <00:06:30.227>of <00:06:30.382>cat <00:06:30.538>pictures, <00:06:30.693>that's <00:06:30.849>my <00:06:31.005>data

124
00:06:33.865 --> 00:06:38.325
<00:06:33.865>and <00:06:33.992>I <00:06:34.119>want <00:06:34.246>to <00:06:34.374>recognize <00:06:34.501>the <00:06:34.628>cat <00:06:34.755>to <00:06:34.883>give <00:06:35.010>the <00:06:35.137>correct <00:06:35.265>bowl

125
00:06:38.385 --> 00:06:40.085
<00:06:38.385>of <00:06:38.664>foods <00:06:38.945>to <00:06:39.224>the <00:06:39.505>correct <00:06:39.784>cat.

126
00:06:42.785 --> 00:06:46.725
<00:06:42.784>But <00:06:42.901>is <00:06:43.018>it <00:06:43.135>all <00:06:43.251>the <00:06:43.368>way, <00:06:43.484>just <00:06:43.601>like <00:06:43.718>I <00:06:43.835>said <00:06:43.951>with <00:06:44.068>the <00:06:44.184>customers

127
00:06:46.725 --> 00:06:47.805
<00:06:46.725>to <00:06:46.887>wanna <00:06:47.049>do <00:06:47.211>it <00:06:47.373>for <00:06:47.535>the <00:06:47.697>hype.

128
00:06:48.235 --> 00:06:51.165
<00:06:48.235>Most <00:06:48.435>of <00:06:48.635>the <00:06:48.835>times <00:06:49.035>the <00:06:49.235>solution <00:06:49.435>is <00:06:49.635>something

129
00:06:51.195 --> 00:06:52.245
<00:06:51.195>with <00:06:51.667>business <00:06:52.140>rules.

130
00:06:52.305 --> 00:06:55.085
<00:06:52.305>If <00:06:52.445>you <00:06:52.585>already <00:06:52.725>have <00:06:52.865>business <00:06:53.005>rules, <00:06:53.145>please <00:06:53.285>do <00:06:53.425>it <00:06:53.565>old <00:06:53.705>way

131
00:06:55.355 --> 00:06:57.605
<00:06:55.355>because <00:06:55.555>you <00:06:55.755>first <00:06:55.955>need <00:06:56.155>a <00:06:56.355>lot <00:06:56.555>of <00:06:56.755>data

132
00:06:58.185 --> 00:07:00.805
<00:06:58.185>and <00:06:58.360>you <00:06:58.535>need <00:06:58.710>an <00:06:58.885>expert <00:06:59.060>to <00:06:59.235>check <00:06:59.410>the <00:06:59.585>data.

133
00:07:01.465 --> 00:07:03.525
<00:07:01.465>So <00:07:01.640>for <00:07:01.815>example, <00:07:01.989>if <00:07:02.164>you're <00:07:02.340>in <00:07:02.515>a <00:07:02.690>medical <00:07:02.864>circuit

134
00:07:04.705 --> 00:07:06.165
<00:07:04.705>and <00:07:04.924>you <00:07:05.143>have <00:07:05.362>a <00:07:05.581>doctor <00:07:05.800>that <00:07:06.019>has

135
00:07:06.165 --> 00:07:09.405
<00:07:06.165>to <00:07:06.320>check <00:07:06.476>if <00:07:06.631>the <00:07:06.787>bone <00:07:06.942>on <00:07:07.098>Emory <00:07:07.253>scan <00:07:07.409>is <00:07:07.565>broken

136
00:07:09.545 --> 00:07:13.165
<00:07:09.545>or <00:07:09.685>not, <00:07:09.825>you <00:07:09.965>want <00:07:10.105>the <00:07:10.245>doctor <00:07:10.385>not <00:07:10.525>to <00:07:10.665>label <00:07:10.805>your <00:07:10.945>data,

137
00:07:13.465 --> 00:07:16.165
<00:07:13.465>but <00:07:13.664>please <00:07:13.864>be <00:07:14.065>operating <00:07:14.265>and <00:07:14.465>helping <00:07:14.664>sick <00:07:14.864>people.

138
00:07:16.785 --> 00:07:20.005
<00:07:16.785>So <00:07:17.065>it's <00:07:17.345>very <00:07:17.625>uh, <00:07:17.905>precious <00:07:18.185>time.

139
00:07:20.065 --> 00:07:23.005
<00:07:20.065>You <00:07:20.240>don't <00:07:20.415>wanna <00:07:20.590>use <00:07:20.765>them <00:07:20.940>for <00:07:21.115>labeling <00:07:21.290>the <00:07:21.465>data,

140
00:07:23.145 --> 00:07:25.045
<00:07:23.145>but <00:07:23.320>you <00:07:23.495>need <00:07:23.669>it <00:07:23.844>because <00:07:24.020>you're <00:07:24.195>not <00:07:24.370>an <00:07:24.544>expert.

141
00:07:26.145 --> 00:07:29.125
<00:07:26.145>And <00:07:26.320>experts <00:07:26.495>most <00:07:26.669>of <00:07:26.844>the <00:07:27.020>time <00:07:27.195>are <00:07:27.370>very <00:07:27.544>expensive.

142
00:07:29.745 --> 00:07:32.285
<00:07:29.745>So <00:07:29.885>this <00:07:30.025>is <00:07:30.165>one <00:07:30.305>of <00:07:30.445>the <00:07:30.585>things <00:07:30.725>I'll <00:07:30.865>say <00:07:31.005>to <00:07:31.145>customers.

143
00:07:32.715 --> 00:07:36.525
<00:07:32.715>It's <00:07:32.890>gonna <00:07:33.065>take <00:07:33.240>a <00:07:33.415>lot <00:07:33.590>of <00:07:33.765>time <00:07:33.940>and <00:07:34.115>money.

144
00:07:38.545 --> 00:07:40.645
<00:07:38.545>Um, <00:07:38.825>let's <00:07:39.105>go <00:07:39.385>to <00:07:39.665>the <00:07:39.945>code.

145
00:07:41.285 --> 00:07:43.125
<00:07:41.284>I <00:07:41.460>also <00:07:41.635>have <00:07:41.809>some <00:07:41.984>life <00:07:42.159>coding <00:07:42.335>on <00:07:42.510>the <00:07:42.684>back,

146
00:07:43.785 --> 00:07:48.125
<00:07:43.784>but <00:07:43.901>I <00:07:44.018>only <00:07:44.135>have <00:07:44.251>22 <00:07:44.368>minutes <00:07:44.484>so <00:07:44.601>I <00:07:44.718>just <00:07:44.835>see <00:07:44.951>it <00:07:45.068>this <00:07:45.184>way.

147
00:07:48.945 --> 00:07:50.885
<00:07:48.945>All <00:07:49.178>the <00:07:49.411>code <00:07:49.645>examples <00:07:49.878>are <00:07:50.111>in <00:07:50.344>Biden.

148
00:07:51.185 --> 00:07:53.245
<00:07:51.185>If <00:07:51.465>you <00:07:51.745>dunno, <00:07:52.025>Biden, <00:07:52.305>no <00:07:52.585>worries.

149
00:07:53.545 --> 00:07:56.005
<00:07:53.545>I'm <00:07:53.720>gonna <00:07:53.895>just <00:07:54.070>go <00:07:54.245>through <00:07:54.420>it <00:07:54.595>step <00:07:54.770>by <00:07:54.945>step.

150
00:07:56.665 --> 00:07:58.605
<00:07:56.665>And <00:07:56.945>afterwards <00:07:57.225>I'm <00:07:57.505>also <00:07:57.785>gonna <00:07:58.065>explain.

151
00:07:59.385 --> 00:08:01.325
<00:07:59.385>So <00:07:59.618>we <00:07:59.851>first <00:08:00.085>do <00:08:00.318>a <00:08:00.551>PIP <00:08:00.784>install.

152
00:08:01.765 --> 00:08:05.245
<00:08:01.765>A <00:08:01.905>PIP <00:08:02.044>install <00:08:02.185>is <00:08:02.325>Python <00:08:02.465>for <00:08:02.604>installing <00:08:02.745>and <00:08:02.885>we <00:08:03.025>have <00:08:03.164>pen

153
00:08:05.385 --> 00:08:08.525
<00:08:05.385>and <00:08:05.735>very <00:08:06.085>common <00:08:06.435>used <00:08:06.784>libraries

154
00:08:09.065 --> 00:08:13.405
<00:08:09.065>and <00:08:09.192>mud <00:08:09.319>plot <00:08:09.446>lip <00:08:09.574>for <00:08:09.701>plotting <00:08:09.828>things <00:08:09.955>we <00:08:10.083>have <00:08:10.210>Sky <00:08:10.337>could <00:08:10.465>learn

155
00:08:13.405 --> 00:08:16.925
<00:08:13.405>because <00:08:13.544>back <00:08:13.684>in <00:08:13.825>the <00:08:13.965>day <00:08:14.104>you <00:08:14.244>needed <00:08:14.385>a <00:08:14.525>lot <00:08:14.664>of <00:08:14.804>mathematics

156
00:08:17.505 --> 00:08:19.405
<00:08:17.505>to <00:08:17.784>understand <00:08:18.065>and <00:08:18.344>to <00:08:18.625>calculate <00:08:18.905>things.

157
00:08:19.745 --> 00:08:24.085
<00:08:19.745>But <00:08:19.900>luckily <00:08:20.056>today <00:08:20.211>you <00:08:20.367>just <00:08:20.522>can <00:08:20.678>call <00:08:20.833>the <00:08:20.989>methods. <00:08:21.145>Anyway.

158
00:08:24.275 --> 00:08:25.885
<00:08:24.275>Jupiter <00:08:24.741>because <00:08:25.208>it's <00:08:25.674>notebook

159
00:08:26.145 --> 00:08:28.165
<00:08:26.145>and <00:08:26.378>Kegel, <00:08:26.611>I'll <00:08:26.844>come <00:08:27.078>on <00:08:27.311>later <00:08:27.544>on

160
00:08:29.025 --> 00:08:33.205
<00:08:29.025>now <00:08:29.200>we <00:08:29.375>import <00:08:29.549>all <00:08:29.724>the <00:08:29.900>different <00:08:30.075>kinds <00:08:30.250>of <00:08:30.424>libraries

161
00:08:35.825 --> 00:08:38.325
<00:08:35.825>and <00:08:36.058>then <00:08:36.291>we <00:08:36.525>gonna <00:08:36.758>collect <00:08:36.991>the <00:08:37.225>data.

162
00:08:39.985 --> 00:08:44.565
<00:08:39.985>So <00:08:40.092>you <00:08:40.200>have <00:08:40.308>to <00:08:40.415>think <00:08:40.523>of <00:08:40.631>where <00:08:40.738>do <00:08:40.846>I <00:08:40.954>want <00:08:41.061>to <00:08:41.169>get <00:08:41.277>my <00:08:41.385>data?

163
00:08:44.595 --> 00:08:45.965
<00:08:44.595>Does <00:08:44.841>the <00:08:45.088>customer <00:08:45.334>have <00:08:45.581>the <00:08:45.828>data?

164
00:08:46.505 --> 00:08:47.725
<00:08:46.505>Do <00:08:46.724>I <00:08:46.944>have <00:08:47.163>my <00:08:47.383>own <00:08:47.603>data

165
00:08:48.505 --> 00:08:51.445
<00:08:48.505>or <00:08:48.660>do <00:08:48.816>I <00:08:48.971>got <00:08:49.127>'em <00:08:49.282>from <00:08:49.438>the <00:08:49.593>web <00:08:49.749>or <00:08:49.905>something?

166
00:08:51.865 --> 00:08:55.245
<00:08:51.865>But <00:08:52.098>then <00:08:52.331>what's <00:08:52.565>are <00:08:52.798>the <00:08:53.031>data <00:08:53.265>requirements?

167
00:08:55.365 --> 00:08:58.085
<00:08:55.365>Because <00:08:55.492>I <00:08:55.619>have <00:08:55.746>all <00:08:55.874>the <00:08:56.001>pictures <00:08:56.128>of <00:08:56.255>my <00:08:56.383>cat <00:08:56.510>on <00:08:56.637>my <00:08:56.765>phone,

168
00:08:58.865 --> 00:09:01.485
<00:08:58.865>but <00:08:59.065>maybe <00:08:59.265>my <00:08:59.465>neighbor <00:08:59.665>also <00:08:59.865>have <00:09:00.065>cat <00:09:00.265>pictures,

169
00:09:01.545 --> 00:09:04.245
<00:09:01.545>but <00:09:01.700>maybe <00:09:01.856>her <00:09:02.011>phone <00:09:02.167>is <00:09:02.322>uh, <00:09:02.478>worse <00:09:02.633>or <00:09:02.789>even <00:09:02.944>better.

170
00:09:04.745 --> 00:09:05.765
<00:09:04.745>So <00:09:04.928>what <00:09:05.112>do <00:09:05.295>we <00:09:05.479>all <00:09:05.663>want?

171
00:09:07.065 --> 00:09:09.605
<00:09:07.065>And <00:09:07.205>when <00:09:07.345>you <00:09:07.485>collected <00:09:07.625>all <00:09:07.765>the <00:09:07.905>data, <00:09:08.045>is <00:09:08.185>it <00:09:08.325>clean <00:09:08.465>enough?

172
00:09:09.605 --> 00:09:13.725
<00:09:09.605>Like <00:09:09.745>aren't <00:09:09.885>there <00:09:10.025>pictures <00:09:10.165>of <00:09:10.305>food <00:09:10.445>in <00:09:10.585>between <00:09:10.725>or <00:09:10.865>from <00:09:11.005>ducks?

173
00:09:14.745 --> 00:09:16.805
<00:09:14.745>And <00:09:14.900>then <00:09:15.056>you <00:09:15.211>need <00:09:15.367>to <00:09:15.522>validate <00:09:15.678>it <00:09:15.833>with <00:09:15.989>the <00:09:16.145>labels

174
00:09:16.805 --> 00:09:19.365
<00:09:16.805>and <00:09:17.038>everything <00:09:17.271>and <00:09:17.505>organize <00:09:17.738>it <00:09:17.971>in <00:09:18.204>place.

175
00:09:21.265 --> 00:09:24.845
<00:09:21.265>But <00:09:21.420>luckily <00:09:21.576>there's <00:09:21.731>already <00:09:21.887>a <00:09:22.042>lot <00:09:22.198>of <00:09:22.353>you <00:09:22.509>for <00:09:22.665>you.

176
00:09:25.545 --> 00:09:29.605
<00:09:25.545>We <00:09:25.685>have <00:09:25.824>uh, <00:09:25.964>for <00:09:26.104>example <00:09:26.245>Kegel, <00:09:26.385>that's <00:09:26.525>a <00:09:26.665>platform <00:09:26.805>to <00:09:26.944>um,

177
00:09:30.715 --> 00:09:33.565
<00:09:30.715>this <00:09:30.890>a <00:09:31.065>competition <00:09:31.240>of <00:09:31.415>who <00:09:31.590>makes <00:09:31.765>the <00:09:31.940>best <00:09:32.115>model.

178
00:09:33.825 --> 00:09:36.045
<00:09:33.825>But <00:09:33.980>there's <00:09:34.136>also <00:09:34.291>a <00:09:34.447>lot <00:09:34.602>of <00:09:34.758>data <00:09:34.913>sets <00:09:35.069>already <00:09:35.225>there.

179
00:09:36.425 --> 00:09:38.285
<00:09:36.425>Google <00:09:36.658>has <00:09:36.891>a <00:09:37.125>lot <00:09:37.358>of <00:09:37.591>data <00:09:37.824>sets

180
00:09:39.105 --> 00:09:42.805
<00:09:39.105>and <00:09:39.260>also <00:09:39.416>the <00:09:39.571>skews <00:09:39.727>also <00:09:39.882>are <00:09:40.038>back <00:09:40.193>at <00:09:40.349>the <00:09:40.505>presentation.

181
00:09:43.505 --> 00:09:45.885
<00:09:43.505>We <00:09:43.705>have <00:09:43.905>the <00:09:44.105>government <00:09:44.305>of <00:09:44.505>the <00:09:44.705>United <00:09:44.905>States

182
00:09:46.345 --> 00:09:47.365
<00:09:46.345>and <00:09:46.651>the <00:09:46.957>Dutch <00:09:47.263>government.

183
00:09:48.385 --> 00:09:50.965
<00:09:48.385>So <00:09:48.560>there's <00:09:48.735>a <00:09:48.910>lot <00:09:49.085>of <00:09:49.260>data <00:09:49.435>sets <00:09:49.610>already <00:09:49.785>there.

184
00:09:51.065 --> 00:09:52.925
<00:09:51.065>So <00:09:51.345>please <00:09:51.625>build <00:09:51.905>up <00:09:52.185>on <00:09:52.465>them

185
00:09:53.275 --> 00:09:55.645
<00:09:53.275>because <00:09:53.475>you <00:09:53.675>don't <00:09:53.875>have <00:09:54.074>to <00:09:54.275>do <00:09:54.475>everything <00:09:54.675>yourself.

186
00:09:58.605 --> 00:10:01.795
<00:09:58.605>Sorry. <00:09:58.885>So <00:09:59.165>back <00:09:59.445>to <00:09:59.725>the <00:10:00.005>codes.

187
00:10:02.835 --> 00:10:04.855
<00:10:02.835>We <00:10:03.115>wanted <00:10:03.395>to <00:10:03.675>connect <00:10:03.955>with <00:10:04.235>Kele

188
00:10:05.395 --> 00:10:09.455
<00:10:05.395>and <00:10:05.550>in <00:10:05.706>the <00:10:05.861>back <00:10:06.017>I <00:10:06.172>had <00:10:06.328>two <00:10:06.483>properties, <00:10:06.639>Kele <00:10:06.795>key

189
00:10:09.595 --> 00:10:11.935
<00:10:09.595>and <00:10:09.945>a <00:10:10.295>kele <00:10:10.645>uh, <00:10:10.995>username

190
00:10:13.005 --> 00:10:15.655
<00:10:13.005>because <00:10:13.132>you <00:10:13.259>have <00:10:13.386>to <00:10:13.514>make <00:10:13.641>an <00:10:13.768>account <00:10:13.895>for <00:10:14.023>it, <00:10:14.150>it's <00:10:14.277>just <00:10:14.405>free.

191
00:10:15.795 --> 00:10:17.135
<00:10:15.795>But <00:10:16.096>we <00:10:16.398>need <00:10:16.699>an <00:10:17.001>account

192
00:10:18.035 --> 00:10:22.135
<00:10:18.035>and <00:10:18.209>we <00:10:18.385>connect <00:10:18.560>with <00:10:18.735>it <00:10:18.910>saying <00:10:19.084>Kele <00:10:19.260>API <00:10:19.435>authenticate

193
00:10:22.885 --> 00:10:27.695
<00:10:22.885>that <00:10:23.012>is <00:10:23.139>and <00:10:23.266>if <00:10:23.394>we <00:10:23.521>load <00:10:23.648>the <00:10:23.775>data <00:10:23.903>sets, <00:10:24.030>I <00:10:24.157>want <00:10:24.285>something

194
00:10:27.695 --> 00:10:28.695
<00:10:27.695>with <00:10:27.995>cats <00:10:28.295>of <00:10:28.595>course.

195
00:10:28.875 --> 00:10:31.815
<00:10:28.875>Uh, <00:10:29.030>so <00:10:29.186>a <00:10:29.341>data <00:10:29.497>set <00:10:29.652>with <00:10:29.808>cat <00:10:29.963>reads <00:10:30.119>in <00:10:30.275>it

196
00:10:33.035 --> 00:10:34.815
<00:10:33.035>and <00:10:33.268>lemme <00:10:33.501>say <00:10:33.735>okay <00:10:33.968>and <00:10:34.201>sip <00:10:34.435>it

197
00:10:34.815 --> 00:10:39.535
<00:10:34.815>because <00:10:34.942>it <00:10:35.069>was <00:10:35.196>zipped <00:10:35.324>in <00:10:35.451>the <00:10:35.578>pot <00:10:35.705>dot <00:10:35.833>because <00:10:35.960>it's <00:10:36.087>at <00:10:36.215>root

198
00:10:40.315 --> 00:10:41.535
<00:10:40.315>and <00:10:40.589>I <00:10:40.864>want <00:10:41.138>the <00:10:41.413>dataset.

199
00:10:42.675 --> 00:10:46.575
<00:10:42.675>And <00:10:42.875>this <00:10:43.074>way <00:10:43.275>you <00:10:43.474>can <00:10:43.675>read <00:10:43.875>the <00:10:44.074>file

200
00:10:48.035 --> 00:10:50.295
<00:10:48.035>and <00:10:48.175>if <00:10:48.314>you <00:10:48.454>look <00:10:48.594>at <00:10:48.735>the <00:10:48.875>file <00:10:49.015>it <00:10:49.155>looks <00:10:49.295>like <00:10:49.435>d***s.

201
00:10:50.755 --> 00:10:53.375
<00:10:50.755>We <00:10:50.955>have <00:10:51.155>links <00:10:51.355>to <00:10:51.555>different <00:10:51.755>kind <00:10:51.955>of <00:10:52.155>pictures.

202
00:10:54.035 --> 00:10:57.775
<00:10:54.035>We <00:10:54.151>have <00:10:54.268>the <00:10:54.385>aids, <00:10:54.501>but <00:10:54.618>if <00:10:54.735>you <00:10:54.851>look <00:10:54.968>closely <00:10:55.084>to <00:10:55.201>it, <00:10:55.318>you <00:10:55.435>see

203
00:10:57.775 --> 00:11:00.055
<00:10:57.775>that's <00:10:57.949>adult <00:10:58.125>and <00:10:58.300>baby <00:10:58.475>and <00:10:58.650>that <00:10:58.824>kind <00:10:59.000>of <00:10:59.175>stuff.

204
00:11:00.115 --> 00:11:03.015
<00:11:00.115>So <00:11:00.315>now <00:11:00.515>days <00:11:00.715>or <00:11:00.915>years. <00:11:01.115>So <00:11:01.315>that's <00:11:01.515>interesting.

205
00:11:03.835 --> 00:11:06.615
<00:11:03.835>And <00:11:03.975>if <00:11:04.115>you <00:11:04.255>see <00:11:04.395>with <00:11:04.535>codes, <00:11:04.675>there's <00:11:04.815>also <00:11:04.955>not <00:11:05.095>a <00:11:05.235>number.

206
00:11:06.795 --> 00:11:08.815
<00:11:06.795>So <00:11:07.074>that's <00:11:07.354>not <00:11:07.635>really <00:11:07.915>clean <00:11:08.194>data.

207
00:11:10.035 --> 00:11:14.015
<00:11:10.035>And <00:11:10.190>the <00:11:10.346>type <00:11:10.501>for <00:11:10.657>the <00:11:10.812>first <00:11:10.968>rows <00:11:11.123>are <00:11:11.279>always <00:11:11.435>cats.

208
00:11:14.275 --> 00:11:16.695
<00:11:14.275>So <00:11:14.508>that's <00:11:14.741>interesting <00:11:14.975>to <00:11:15.208>take <00:11:15.441>a <00:11:15.675>mind.

209
00:11:17.555 --> 00:11:18.615
<00:11:17.555>And <00:11:17.793>we <00:11:18.031>can <00:11:18.270>see <00:11:18.509>this

210
00:11:18.755 --> 00:11:21.615
<00:11:18.755>by <00:11:18.910>calling <00:11:19.066>the <00:11:19.221>data <00:11:19.377>frame <00:11:19.532>the <00:11:19.688>cat <00:11:19.843>on <00:11:19.999>square <00:11:20.155>data

211
00:11:21.615 --> 00:11:22.895
<00:11:21.615>frame <00:11:22.191>dot <00:11:22.766>hat.

212
00:11:24.845 --> 00:11:29.105
<00:11:24.845>So <00:11:25.078>we <00:11:25.311>have <00:11:25.545>to <00:11:25.778>do <00:11:26.011>some <00:11:26.245>pre-processing,

213
00:11:30.165 --> 00:11:31.945
<00:11:30.165>we <00:11:30.444>need <00:11:30.724>to <00:11:31.005>clean <00:11:31.285>the <00:11:31.564>data

214
00:11:32.885 --> 00:11:35.665
<00:11:32.885>and <00:11:33.001>then <00:11:33.118>we <00:11:33.235>transform <00:11:33.351>it <00:11:33.468>to <00:11:33.585>do <00:11:33.701>it <00:11:33.818>all <00:11:33.935>in <00:11:34.051>the <00:11:34.168>same <00:11:34.285>format.

215
00:11:36.935 --> 00:11:41.185
<00:11:36.935>Some <00:11:37.214>extractions, <00:11:37.494>um, <00:11:37.775>from <00:11:38.055>features <00:11:38.334>like

216
00:11:41.245 --> 00:11:43.985
<00:11:41.245>that's <00:11:41.385>already <00:11:41.525>in <00:11:41.665>the <00:11:41.805>CSV <00:11:41.945>in <00:11:42.085>this <00:11:42.225>file, <00:11:42.365>like <00:11:42.505>the <00:11:42.645>aids

217
00:11:44.365 --> 00:11:46.305
<00:11:44.365>and <00:11:44.540>the <00:11:44.715>gender <00:11:44.890>and <00:11:45.065>all <00:11:45.240>that <00:11:45.415>kind <00:11:45.590>of <00:11:45.765>stuff.

218
00:11:47.755 --> 00:11:51.815
<00:11:47.755>But <00:11:47.910>you <00:11:48.066>can <00:11:48.221>also <00:11:48.377>extract <00:11:48.532>new <00:11:48.688>features <00:11:48.843>that's <00:11:48.999>already <00:11:49.155>there.

219
00:11:51.915 --> 00:11:53.205
<00:11:51.915>For <00:11:52.205>example, <00:11:52.495>all <00:11:52.785>the <00:11:53.076>cats

220
00:11:53.205 --> 00:11:55.805
<00:11:53.205>with <00:11:53.438>blue <00:11:53.671>eyes <00:11:53.905>are <00:11:54.138>babies <00:11:54.371>or <00:11:54.605>something.

221
00:11:55.945 --> 00:11:58.125
<00:11:55.945>So <00:11:56.178>that's <00:11:56.411>really <00:11:56.645>something <00:11:56.878>you <00:11:57.111>can <00:11:57.345>use

222
00:12:00.065 --> 00:12:03.445
<00:12:00.065>and <00:12:00.181>we <00:12:00.298>need <00:12:00.415>to <00:12:00.531>split <00:12:00.648>the <00:12:00.765>data, <00:12:00.881>but <00:12:00.998>I'll <00:12:01.115>come <00:12:01.231>there <00:12:01.348>on <00:12:01.465>later.

223
00:12:05.085 --> 00:12:09.985
<00:12:05.085>So <00:12:05.212>back <00:12:05.339>to <00:12:05.466>the <00:12:05.594>code <00:12:05.721>we <00:12:05.848>have <00:12:05.975>here <00:12:06.103>that <00:12:06.230>we <00:12:06.357>say, <00:12:06.485>okay,

224
00:12:10.295 --> 00:12:12.785
<00:12:10.295>what <00:12:10.761>columns <00:12:11.228>are <00:12:11.694>relevant?

225
00:12:13.325 --> 00:12:16.625
<00:12:13.325>So <00:12:13.441>we <00:12:13.558>have <00:12:13.675>the <00:12:13.791>age <00:12:13.908>and <00:12:14.025>the <00:12:14.141>gender, <00:12:14.258>the <00:12:14.375>code <00:12:14.491>and <00:12:14.608>the <00:12:14.725>size

226
00:12:17.405 --> 00:12:20.305
<00:12:17.405>and <00:12:17.755>everything <00:12:18.105>with <00:12:18.454>missing <00:12:18.805>value.

227
00:12:20.445 --> 00:12:22.825
<00:12:20.445>So <00:12:20.645>another <00:12:20.845>numbers <00:12:21.045>for <00:12:21.245>example, <00:12:21.445>we're <00:12:21.645>gonna <00:12:21.845>drop,

228
00:12:22.885 --> 00:12:24.705
<00:12:22.885>we <00:12:23.118>don't <00:12:23.351>need <00:12:23.585>them, <00:12:23.818>it's <00:12:24.051>incomplete <00:12:24.285>data,

229
00:12:25.015 --> 00:12:26.345
<00:12:25.015>it's <00:12:25.214>not <00:12:25.414>very <00:12:25.613>good <00:12:25.813>for <00:12:26.012>our <00:12:26.212>model.

230
00:12:27.775 --> 00:12:32.665
<00:12:27.775>Then <00:12:27.915>we <00:12:28.055>say <00:12:28.194>okay, <00:12:28.334>in <00:12:28.475>machine <00:12:28.615>learning <00:12:28.755>we <00:12:28.895>always <00:12:29.035>use <00:12:29.175>X

231
00:12:32.805 --> 00:12:37.025
<00:12:32.805>and <00:12:32.921>Y <00:12:33.038>and <00:12:33.155>x <00:12:33.271>we're <00:12:33.388>doing <00:12:33.505>the <00:12:33.621>stuff <00:12:33.738>we <00:12:33.854>wanna <00:12:33.971>predict <00:12:34.088>it <00:12:34.204>with.

232
00:12:37.125 --> 00:12:38.865
<00:12:37.125>So <00:12:37.475>the <00:12:37.825>H <00:12:38.175>gener <00:12:38.525>codes

233
00:12:39.245 --> 00:12:42.065
<00:12:39.245>and <00:12:39.361>the <00:12:39.478>Y, <00:12:39.595>what <00:12:39.711>we <00:12:39.828>wanna <00:12:39.945>predict, <00:12:40.061>so <00:12:40.178>the <00:12:40.295>size <00:12:40.411>of <00:12:40.528>the <00:12:40.645>cat

234
00:12:43.965 --> 00:12:47.785
<00:12:43.965>and <00:12:44.120>then <00:12:44.276>we <00:12:44.431>use <00:12:44.587>something <00:12:44.742>that's <00:12:44.898>called <00:12:45.053>the <00:12:45.209>label <00:12:45.365>encoder.

235
00:12:48.045 --> 00:12:52.145
<00:12:48.045>And <00:12:48.219>in <00:12:48.395>the <00:12:48.569>label <00:12:48.745>encoder <00:12:48.920>we're <00:12:49.094>doing <00:12:49.270>the <00:12:49.444>mapping

236
00:12:52.325 --> 00:12:53.505
<00:12:52.325>to <00:12:52.856>numerical <00:12:53.387>failures.

237
00:12:53.525 --> 00:12:54.665
<00:12:53.525>So <00:12:53.781>the <00:12:54.038>model <00:12:54.294>is <00:12:54.550>faster

238
00:12:54.665 --> 00:12:59.385
<00:12:54.665>because <00:12:54.792>what <00:12:54.919>you <00:12:55.046>saw <00:12:55.174>with <00:12:55.301>the <00:12:55.428>age <00:12:55.555>that's <00:12:55.683>uh, <00:12:55.810>it <00:12:55.937>was <00:12:56.064>young,

239
00:12:59.415 --> 00:13:01.145
<00:12:59.415>yeah, <00:12:59.694>he <00:12:59.974>doesn't <00:13:00.255>understand <00:13:00.535>the <00:13:00.814>model.

240
00:13:02.005 --> 00:13:03.825
<00:13:02.005>So <00:13:02.238>that's <00:13:02.471>why <00:13:02.705>we <00:13:02.938>use <00:13:03.171>label <00:13:03.405>encoder

241
00:13:05.685 --> 00:13:08.425
<00:13:05.685>and <00:13:05.918>we <00:13:06.151>use <00:13:06.385>the <00:13:06.618>one <00:13:06.851>hot <00:13:07.084>encoder

242
00:13:09.765 --> 00:13:12.425
<00:13:09.765>and <00:13:10.045>that's <00:13:10.324>about <00:13:10.605>combining <00:13:10.885>those <00:13:11.165>two.

243
00:13:12.685 --> 00:13:16.425
<00:13:12.685>So <00:13:12.824>the <00:13:12.964>H <00:13:13.104>general <00:13:13.244>code <00:13:13.385>making <00:13:13.525>it <00:13:13.665>one <00:13:13.805>big <00:13:13.944>failure <00:13:14.084>also

244
00:13:16.485 --> 00:13:17.985
<00:13:16.485>for <00:13:17.835>performance.

245
00:13:20.895 --> 00:13:24.995
<00:13:20.895>So <00:13:21.035>we <00:13:21.175>have <00:13:21.314>this <00:13:21.454>and <00:13:21.595>then <00:13:21.735>we're <00:13:21.875>going <00:13:22.015>to <00:13:22.155>data <00:13:22.295>process

246
00:13:27.055 --> 00:13:30.395
<00:13:27.055>And <00:13:27.210>there <00:13:27.366>is <00:13:27.521>a <00:13:27.677>case, <00:13:27.832>you <00:13:27.988>don't <00:13:28.143>have <00:13:28.299>enough <00:13:28.454>data

247
00:13:30.935 --> 00:13:34.195
<00:13:30.935>and <00:13:31.074>if <00:13:31.214>you <00:13:31.354>don't <00:13:31.494>have <00:13:31.635>enough <00:13:31.775>data <00:13:31.915>you <00:13:32.055>can <00:13:32.194>create <00:13:32.334>more.

248
00:13:34.615 --> 00:13:36.635
<00:13:34.615>And <00:13:34.965>it's <00:13:35.315>called <00:13:35.665>data <00:13:36.015>augmentation.

249
00:13:37.135 --> 00:13:39.435
<00:13:37.135>So <00:13:37.335>you <00:13:37.535>see <00:13:37.735>here <00:13:37.935>a <00:13:38.135>very <00:13:38.335>cute <00:13:38.535>cat

250
00:13:40.415 --> 00:13:42.275
<00:13:40.415>and <00:13:40.765>we <00:13:41.115>take <00:13:41.464>the <00:13:41.814>picture

251
00:13:42.655 --> 00:13:47.035
<00:13:42.655>and <00:13:42.782>we <00:13:42.909>changed <00:13:43.036>a <00:13:43.164>little <00:13:43.291>bit, <00:13:43.418>play <00:13:43.545>with <00:13:43.673>RGB <00:13:43.800>values, <00:13:43.927>zoom <00:13:44.055>in,

252
00:13:47.185 --> 00:13:50.355
<00:13:47.185>zoom <00:13:47.418>out <00:13:47.651>a <00:13:47.885>bam <00:13:48.118>six <00:13:48.351>new <00:13:48.584>pictures.

253
00:13:50.815 --> 00:13:51.915
<00:13:50.815>So <00:13:51.013>this <00:13:51.211>is <00:13:51.409>a <00:13:51.607>great <00:13:51.805>way

254
00:13:52.255 --> 00:13:55.755
<00:13:52.255>to <00:13:52.430>create <00:13:52.605>more <00:13:52.780>data <00:13:52.955>in <00:13:53.130>a <00:13:53.305>very <00:13:53.480>short <00:13:53.655>time.

255
00:13:57.015 --> 00:14:01.075
<00:13:57.015>But <00:13:57.155>if <00:13:57.295>you're <00:13:57.435>doing <00:13:57.574>this, <00:13:57.715>there's <00:13:57.855>a <00:13:57.995>chance <00:13:58.135>you <00:13:58.275>can <00:13:58.415>go

256
00:14:01.585 --> 00:14:03.835
<00:14:01.585>overfit <00:14:02.285>and <00:14:02.985>overfit.

257
00:14:04.335 --> 00:14:07.635
<00:14:04.335>That's <00:14:04.490>like <00:14:04.646>if <00:14:04.801>you <00:14:04.957>only <00:14:05.112>train <00:14:05.268>your <00:14:05.423>model <00:14:05.579>with <00:14:05.735>pictures

258
00:14:07.735 --> 00:14:09.035
<00:14:07.735>of <00:14:08.027>cats <00:14:08.319>on <00:14:08.612>the <00:14:08.905>grass,

259
00:14:10.145 --> 00:14:12.515
<00:14:10.145>then <00:14:10.425>your <00:14:10.704>model <00:14:10.985>will <00:14:11.265>only <00:14:11.545>recognize

260
00:14:12.545 --> 00:14:13.955
<00:14:12.545>cats <00:14:12.798>if <00:14:13.052>they're <00:14:13.306>on <00:14:13.560>the <00:14:13.814>grass.

261
00:14:14.135 --> 00:14:16.515
<00:14:14.135>So <00:14:14.251>if <00:14:14.368>you <00:14:14.485>give <00:14:14.601>it <00:14:14.718>a <00:14:14.835>picture <00:14:14.951>of <00:14:15.068>a <00:14:15.185>cat <00:14:15.301>on <00:14:15.418>the <00:14:15.535>street

262
00:14:16.515 --> 00:14:17.595
<00:14:16.515>that <00:14:16.653>says, <00:14:16.792>yeah, <00:14:16.931>this <00:14:17.070>is <00:14:17.209>not <00:14:17.348>a <00:14:17.487>cat.

263
00:14:18.335 --> 00:14:21.755
<00:14:18.335>So <00:14:18.510>that's <00:14:18.685>something <00:14:18.860>you <00:14:19.035>have <00:14:19.210>to <00:14:19.385>take <00:14:19.560>a <00:14:19.735>mines,

264
00:14:25.345 --> 00:14:28.675
<00:14:25.345>Okay, <00:14:25.461>now <00:14:25.578>we <00:14:25.695>have <00:14:25.811>all <00:14:25.928>the <00:14:26.045>data <00:14:26.161>and <00:14:26.278>now <00:14:26.395>we're <00:14:26.511>gonna <00:14:26.628>split <00:14:26.745>it.

265
00:14:29.215 --> 00:14:30.915
<00:14:29.215>So <00:14:29.415>the <00:14:29.615>data <00:14:29.815>set <00:14:30.015>is <00:14:30.215>all <00:14:30.415>our <00:14:30.615>data

266
00:14:31.015 --> 00:14:35.035
<00:14:31.015>and <00:14:31.189>we <00:14:31.365>split <00:14:31.540>it <00:14:31.715>in <00:14:31.890>training, <00:14:32.064>validation <00:14:32.240>and <00:14:32.415>testing.

267
00:14:36.335 --> 00:14:38.035
<00:14:36.335>So <00:14:36.615>it's <00:14:36.895>just <00:14:37.175>like <00:14:37.455>high <00:14:37.735>school.

268
00:14:38.045 --> 00:14:41.315
<00:14:38.045>Again, <00:14:38.245>training <00:14:38.444>is <00:14:38.645>just <00:14:38.844>when <00:14:39.045>you <00:14:39.245>learn <00:14:39.444>something

269
00:14:42.215 --> 00:14:44.355
<00:14:42.215>and <00:14:42.415>when <00:14:42.615>you're <00:14:42.815>thinking <00:14:43.015>this <00:14:43.215>is <00:14:43.415>good <00:14:43.615>enough,

270
00:14:44.775 --> 00:14:46.195
<00:14:44.775>you <00:14:45.094>go <00:14:45.414>into <00:14:45.733>the <00:14:46.053>validation

271
00:14:46.655 --> 00:14:48.195
<00:14:46.655>and <00:14:47.001>the <00:14:47.348>validation <00:14:47.694>is <00:14:48.041>like

272
00:14:48.795 --> 00:14:50.555
<00:14:48.795>a <00:14:49.028>little <00:14:49.261>test, <00:14:49.495>is <00:14:49.728>this <00:14:49.961>good <00:14:50.194>enough?

273
00:14:51.425 --> 00:14:53.605
<00:14:51.425>If <00:14:51.625>not, <00:14:51.824>let's <00:14:52.025>go <00:14:52.224>back <00:14:52.425>to <00:14:52.625>the <00:14:52.824>training.

274
00:14:54.185 --> 00:14:58.085
<00:14:54.185>If <00:14:54.292>so, <00:14:54.400>you <00:14:54.508>can <00:14:54.615>go <00:14:54.723>on <00:14:54.831>and <00:14:54.938>you <00:14:55.046>can <00:14:55.154>go <00:14:55.261>to <00:14:55.369>the <00:14:55.477>testing <00:14:55.584>set

275
00:14:58.425 --> 00:15:01.485
<00:14:58.425>and <00:14:58.580>in <00:14:58.736>the <00:14:58.891>testing <00:14:59.047>set <00:14:59.202>it's <00:14:59.358>like <00:14:59.513>the <00:14:59.669>final <00:14:59.824>exome.

276
00:15:01.665 --> 00:15:04.405
<00:15:01.665>Is <00:15:01.865>this <00:15:02.064>good <00:15:02.265>enough <00:15:02.464>for <00:15:02.665>the <00:15:02.865>real <00:15:03.064>world?

277
00:15:06.065 --> 00:15:07.685
<00:15:06.065>And <00:15:06.345>maybe <00:15:06.625>you <00:15:06.905>think, <00:15:07.185>okay <00:15:07.465>fine,

278
00:15:07.865 --> 00:15:10.085
<00:15:07.865>but <00:15:08.098>how <00:15:08.331>are <00:15:08.565>the <00:15:08.798>proportions <00:15:09.031>of <00:15:09.265>that?

279
00:15:12.025 --> 00:15:16.245
<00:15:12.025>So <00:15:12.152>most <00:15:12.279>of <00:15:12.406>the <00:15:12.534>time <00:15:12.661>we <00:15:12.788>do <00:15:12.915>80% <00:15:13.043>for <00:15:13.170>the <00:15:13.297>training <00:15:13.425>data,

280
00:15:16.825 --> 00:15:20.805
<00:15:16.825>10% <00:15:17.000>for <00:15:17.175>the <00:15:17.350>testing <00:15:17.525>and <00:15:17.700>10% <00:15:17.875>for <00:15:18.050>the <00:15:18.225>validation.

281
00:15:21.745 --> 00:15:24.005
<00:15:21.745>And <00:15:21.978>the <00:15:22.211>testing <00:15:22.445>and <00:15:22.678>the <00:15:22.911>validations <00:15:23.145>are

282
00:15:24.005 --> 00:15:25.165
<00:15:24.005>always <00:15:24.353>the <00:15:24.701>same <00:15:25.049>size.

283
00:15:26.065 --> 00:15:29.725
<00:15:26.065>How <00:15:26.192>more <00:15:26.319>data <00:15:26.446>you <00:15:26.574>have, <00:15:26.701>you <00:15:26.828>can <00:15:26.955>also <00:15:27.083>go <00:15:27.210>from <00:15:27.337>80% <00:15:27.465>training

284
00:15:30.065 --> 00:15:33.845
<00:15:30.065>to <00:15:30.240>70 <00:15:30.415>or <00:15:30.590>even <00:15:30.765>60% <00:15:30.940>of <00:15:31.115>the <00:15:31.290>training <00:15:31.465>data.

285
00:15:35.585 --> 00:15:37.525
<00:15:35.585>So <00:15:35.818>how <00:15:36.051>does <00:15:36.285>that <00:15:36.518>look <00:15:36.751>in <00:15:36.985>codes

286
00:15:38.355 --> 00:15:39.845
<00:15:38.355>also <00:15:38.578>just <00:15:38.802>a <00:15:39.025>method <00:15:39.249>for <00:15:39.472>it <00:15:39.696>luckily.

287
00:15:41.105 --> 00:15:43.085
<00:15:41.105>So <00:15:41.338>we <00:15:41.571>call <00:15:41.805>it <00:15:42.038>train <00:15:42.271>test <00:15:42.505>splits

288
00:15:43.585 --> 00:15:45.805
<00:15:43.585>and <00:15:43.785>we <00:15:43.985>give <00:15:44.185>it <00:15:44.385>something <00:15:44.585>the <00:15:44.785>the <00:15:44.985>X

289
00:15:45.805 --> 00:15:48.005
<00:15:45.805>and <00:15:45.944>the <00:15:46.084>Y, <00:15:46.224>what <00:15:46.364>we <00:15:46.505>want, <00:15:46.645>what <00:15:46.785>we <00:15:46.925>want <00:15:47.064>to <00:15:47.204>predict

290
00:15:48.785 --> 00:15:52.405
<00:15:48.785>the <00:15:48.959>test <00:15:49.135>size <00:15:49.310>of <00:15:49.485>a <00:15:49.660>0.2, <00:15:49.834>that <00:15:50.010>means <00:15:50.185>20%

291
00:15:52.425 --> 00:15:53.885
<00:15:52.425>of <00:15:52.863>our <00:15:53.300>test <00:15:53.739>sets,

292
00:15:54.305 --> 00:15:57.645
<00:15:54.305>20% <00:15:54.505>from <00:15:54.704>a <00:15:54.905>dataset <00:15:55.104>is <00:15:55.305>our <00:15:55.505>test <00:15:55.704>set.

293
00:15:58.945 --> 00:16:00.485
<00:15:58.945>And <00:15:59.222>I <00:15:59.499>use <00:15:59.776>here <00:16:00.053>random <00:16:00.331>state

294
00:16:00.485 --> 00:16:02.765
<00:16:00.485>because <00:16:00.685>then <00:16:00.885>for <00:16:01.085>demos <00:16:01.285>it's <00:16:01.485>always <00:16:01.685>the <00:16:01.885>same.

295
00:16:04.465 --> 00:16:06.605
<00:16:04.465>So <00:16:04.620>here <00:16:04.776>you <00:16:04.931>see <00:16:05.087>we <00:16:05.242>only <00:16:05.398>use <00:16:05.553>a <00:16:05.709>test <00:16:05.865>set

296
00:16:06.605 --> 00:16:08.805
<00:16:06.605>that's <00:16:06.885>also <00:16:07.165>a <00:16:07.445>valid <00:16:07.725>S <00:16:08.005>option.

297
00:16:08.865 --> 00:16:10.165
<00:16:08.865>So <00:16:09.157>you <00:16:09.450>can <00:16:09.742>skip <00:16:10.035>the

298
00:16:10.165 --> 00:16:14.425
<00:16:10.165>validation <00:16:11.564>algorithms.

299
00:16:14.595 --> 00:16:16.105
<00:16:14.595>There <00:16:14.866>are <00:16:15.138>a <00:16:15.410>lot <00:16:15.682>of <00:16:15.954>algorithms.

300
00:16:16.265 --> 00:16:17.705
<00:16:16.265>I <00:16:16.481>can <00:16:16.697>talk <00:16:16.913>about <00:16:17.129>days <00:16:17.345>for <00:16:17.561>them,

301
00:16:18.125 --> 00:16:21.305
<00:16:18.125>but <00:16:18.280>I'll <00:16:18.436>just <00:16:18.591>talk <00:16:18.747>about <00:16:18.902>most <00:16:19.058>common <00:16:19.213>ones <00:16:19.369>for <00:16:19.525>now.

302
00:16:23.405 --> 00:16:24.735
<00:16:23.405>Just <00:16:23.644>taking <00:16:23.883>a <00:16:24.123>sip <00:16:24.362>of <00:16:24.602>water.

303
00:16:30.275 --> 00:16:33.535
<00:16:30.275>So <00:16:30.475>the <00:16:30.675>first <00:16:30.875>one <00:16:31.074>is <00:16:31.275>the <00:16:31.475>linear <00:16:31.675>regression

304
00:16:33.555 --> 00:16:37.135
<00:16:33.555>and <00:16:33.682>if <00:16:33.809>you <00:16:33.936>can <00:16:34.064>see <00:16:34.191>here, <00:16:34.318>it <00:16:34.445>has <00:16:34.573>a <00:16:34.700>line <00:16:34.827>in <00:16:34.954>it,

305
00:16:37.285 --> 00:16:39.695
<00:16:37.285>different <00:16:37.635>different <00:16:37.985>data <00:16:38.334>points <00:16:38.685>and,

306
00:16:39.695 --> 00:16:42.015
<00:16:39.695>and <00:16:39.928>we <00:16:40.161>wanna <00:16:40.395>gets <00:16:40.628>the <00:16:40.861>optimized <00:16:41.095>line

307
00:16:42.525 --> 00:16:44.015
<00:16:42.525>between <00:16:42.972>those <00:16:43.419>data <00:16:43.866>points.

308
00:16:45.315 --> 00:16:47.255
<00:16:45.315>And <00:16:45.548>if <00:16:45.781>you <00:16:46.015>translate <00:16:46.248>that <00:16:46.481>to <00:16:46.715>cats,

309
00:16:47.955 --> 00:16:51.735
<00:16:47.955>we <00:16:48.110>have <00:16:48.266>the <00:16:48.421>weight <00:16:48.577>in <00:16:48.732>grams <00:16:48.888>to <00:16:49.043>the <00:16:49.199>aging <00:16:49.355>days,

310
00:16:52.475 --> 00:16:56.255
<00:16:52.475>we <00:16:52.630>have <00:16:52.786>different <00:16:52.941>data <00:16:53.097>points <00:16:53.252>and <00:16:53.408>then <00:16:53.563>that's <00:16:53.719>the <00:16:53.875>line.

311
00:16:56.675 --> 00:17:00.735
<00:16:56.675>And <00:16:56.908>it's <00:16:57.141>mathematic <00:16:57.375>mathematics, <00:16:57.608>sorry, <00:16:57.841>it's <00:16:58.074>always

312
00:17:01.375 --> 00:17:02.935
<00:17:01.375>Y <00:17:01.655>is <00:17:01.935>A <00:17:02.215>x <00:17:02.495>plus <00:17:02.775>B.

313
00:17:04.315 --> 00:17:05.735
<00:17:04.315>So <00:17:04.570>it's <00:17:04.826>always <00:17:05.081>from <00:17:05.337>the <00:17:05.592>left

314
00:17:06.275 --> 00:17:08.455
<00:17:06.275>and <00:17:06.450>the <00:17:06.625>corner <00:17:06.800>to <00:17:06.975>the <00:17:07.150>right <00:17:07.325>at <00:17:07.500>the <00:17:07.675>corner.

315
00:17:10.285 --> 00:17:12.575
<00:17:10.285>Then <00:17:10.635>we <00:17:10.985>have <00:17:11.335>logistic <00:17:11.685>regression.

316
00:17:13.115 --> 00:17:16.455
<00:17:13.115>And <00:17:13.348>logistic <00:17:13.581>regression <00:17:13.815>is <00:17:14.048>always <00:17:14.281>between <00:17:14.515>zero

317
00:17:16.675 --> 00:17:19.335
<00:17:16.675>and <00:17:16.875>one <00:17:17.075>most <00:17:17.274>of <00:17:17.474>the <00:17:17.675>time <00:17:17.875>for <00:17:18.075>classification.

318
00:17:20.635 --> 00:17:25.535
<00:17:20.635>So <00:17:20.790>it's <00:17:20.946>get <00:17:21.101>picture <00:17:21.257>ordained <00:17:21.412>and <00:17:21.568>then <00:17:21.723>in <00:17:21.879>a <00:17:22.035>time,

319
00:17:25.635 --> 00:17:28.655
<00:17:25.635>and <00:17:25.810>if <00:17:25.984>you <00:17:26.160>see <00:17:26.335>it's <00:17:26.510>like <00:17:26.685>an <00:17:26.859>S <00:17:27.035>form

320
00:17:29.075 --> 00:17:31.655
<00:17:29.075>and <00:17:29.275>it's <00:17:29.475>like <00:17:29.675>a <00:17:29.875>sigmoid <00:17:30.075>or <00:17:30.275>a <00:17:30.475>sinus

321
00:17:33.275 --> 00:17:35.575
<00:17:33.275>and <00:17:33.450>you <00:17:33.625>probably <00:17:33.800>all <00:17:33.975>know <00:17:34.150>this, <00:17:34.325>the <00:17:34.500>decision <00:17:34.675>tree

322
00:17:36.155 --> 00:17:39.575
<00:17:36.155>and <00:17:36.355>decision <00:17:36.555>tree <00:17:36.754>is <00:17:36.955>more <00:17:37.155>about <00:17:37.355>different <00:17:37.555>features.

323
00:17:39.755 --> 00:17:42.215
<00:17:39.755>The <00:17:39.988>most <00:17:40.221>weighted <00:17:40.455>features <00:17:40.688>is <00:17:40.921>on <00:17:41.155>top.

324
00:17:42.715 --> 00:17:45.935
<00:17:42.715>And <00:17:42.889>then <00:17:43.064>based <00:17:43.240>on <00:17:43.415>the <00:17:43.590>other <00:17:43.764>features, <00:17:43.939>it's <00:17:44.115>going

325
00:17:45.935 --> 00:17:48.455
<00:17:45.935>to <00:17:46.168>make <00:17:46.401>a <00:17:46.635>tree <00:17:46.868>getting <00:17:47.101>every <00:17:47.335>leaf.

326
00:17:48.795 --> 00:17:53.285
<00:17:48.795>And <00:17:48.922>then <00:17:49.049>based <00:17:49.176>on <00:17:49.304>a <00:17:49.431>path, <00:17:49.558>it's <00:17:49.685>gonna <00:17:49.813>give <00:17:49.940>it <00:17:50.067>a <00:17:50.195>category.

327
00:17:56.625 --> 00:17:59.365
<00:17:56.625>And <00:17:56.858>the <00:17:57.091>random <00:17:57.325>forest <00:17:57.558>regression, <00:17:57.791>that's <00:17:58.025>like

328
00:18:00.225 --> 00:18:03.765
<00:18:00.225>the <00:18:00.458>decision <00:18:00.691>tree <00:18:00.925>times <00:18:01.158>whatever <00:18:01.391>you <00:18:01.625>want.

329
00:18:04.345 --> 00:18:08.485
<00:18:04.345>And <00:18:04.520>it <00:18:04.695>always <00:18:04.870>choose <00:18:05.045>the <00:18:05.220>more <00:18:05.395>popular <00:18:05.570>uh, <00:18:05.745>opinion.

330
00:18:09.385 --> 00:18:11.365
<00:18:09.385>So <00:18:09.618>if <00:18:09.851>you <00:18:10.085>do <00:18:10.318>three <00:18:10.551>decision <00:18:10.785>trees

331
00:18:11.945 --> 00:18:13.685
<00:18:11.945>and <00:18:12.145>two <00:18:12.345>times <00:18:12.544>it <00:18:12.745>says <00:18:12.945>it's <00:18:13.145>a <00:18:13.345>duck

332
00:18:14.025 --> 00:18:15.565
<00:18:14.025>and <00:18:14.256>one <00:18:14.487>time <00:18:14.718>it <00:18:14.949>says <00:18:15.180>it's <00:18:15.411>cat,

333
00:18:16.025 --> 00:18:17.925
<00:18:16.025>the <00:18:16.305>majority <00:18:16.585>says <00:18:16.865>it's <00:18:17.145>a <00:18:17.425>duck.

334
00:18:18.145 --> 00:18:22.045
<00:18:18.145>So <00:18:18.261>it <00:18:18.378>is <00:18:18.495>a <00:18:18.611>duck. <00:18:18.728>So <00:18:18.845>let's <00:18:18.961>go <00:18:19.078>back <00:18:19.195>to <00:18:19.311>the <00:18:19.428>code <00:18:19.545>again.

335
00:18:24.545 --> 00:18:26.435
<00:18:24.545>Here <00:18:24.895>we <00:18:25.245>see <00:18:25.595>the <00:18:25.945>predecessor.

336
00:18:26.495 --> 00:18:28.555
<00:18:26.495>I'm <00:18:26.774>using <00:18:27.054>the <00:18:27.334>random <00:18:27.614>force <00:18:27.895>classifier.

337
00:18:28.575 --> 00:18:33.235
<00:18:28.575>So <00:18:28.750>the <00:18:28.925>one <00:18:29.100>we <00:18:29.275>just <00:18:29.450>saw, <00:18:29.625>MCA <00:18:29.800>model <00:18:29.975>fit

338
00:18:33.585 --> 00:18:38.315
<00:18:33.585>with <00:18:33.725>X <00:18:33.865>and <00:18:34.005>Y <00:18:34.145>train <00:18:34.285>Easy <00:18:34.425>is <00:18:34.565>that <00:18:34.705>and <00:18:34.845>then <00:18:34.985>predicts

339
00:18:39.395 --> 00:18:42.175
<00:18:39.395>and <00:18:39.628>it's <00:18:39.861>been <00:18:40.095>already <00:18:40.328>done <00:18:40.561>nothing, <00:18:40.795>uh,

340
00:18:44.115 --> 00:18:45.495
<00:18:44.115>to <00:18:45.357>hard.

341
00:18:46.315 --> 00:18:48.895
<00:18:46.315>But <00:18:46.515>then <00:18:46.715>you're <00:18:46.915>probably <00:18:47.115>okay, <00:18:47.315>now <00:18:47.515>train <00:18:47.715>it,

342
00:18:48.915 --> 00:18:51.135
<00:18:48.915>but <00:18:49.090>I <00:18:49.264>don't <00:18:49.440>see <00:18:49.615>anything, <00:18:49.790>what <00:18:49.965>do <00:18:50.139>I <00:18:50.315>want?

343
00:18:52.375 --> 00:18:55.145
<00:18:52.375>Therefore <00:18:52.725>we <00:18:53.075>have <00:18:53.425>model <00:18:53.775>evaluation.

344
00:18:57.245 --> 00:19:01.065
<00:18:57.245>And <00:18:57.419>like <00:18:57.594>you <00:18:57.770>see <00:18:57.945>here, <00:18:58.120>uh, <00:18:58.294>on <00:18:58.469>the <00:18:58.645>top

345
00:19:01.605 --> 00:19:03.025
<00:19:01.605>we <00:19:01.860>see <00:19:02.116>the <00:19:02.371>over <00:19:02.627>fitting <00:19:02.883>problem.

346
00:19:03.205 --> 00:19:08.105
<00:19:03.205>So <00:19:03.332>trying <00:19:03.459>on <00:19:03.586>the <00:19:03.714>top <00:19:03.841>left, <00:19:03.968>trying <00:19:04.095>to <00:19:04.223>get <00:19:04.350>every <00:19:04.477>data <00:19:04.605>point

347
00:19:08.165 --> 00:19:10.985
<00:19:08.165>and <00:19:08.398>a <00:19:08.631>correct <00:19:08.865>category, <00:19:09.098>please <00:19:09.331>don't <00:19:09.565>know

348
00:19:10.985 --> 00:19:13.825
<00:19:10.985>that <00:19:11.140>it's <00:19:11.296>okay <00:19:11.451>if <00:19:11.607>some <00:19:11.762>of <00:19:11.918>the <00:19:12.073>data <00:19:12.229>points <00:19:12.385>are

349
00:19:13.855 --> 00:19:15.985
<00:19:13.855>outside <00:19:14.205>the <00:19:14.555>category <00:19:14.905>they're <00:19:15.255>into.

350
00:19:16.855 --> 00:19:20.145
<00:19:16.855>Because <00:19:17.030>in <00:19:17.205>the <00:19:17.380>real <00:19:17.555>life <00:19:17.730>it's <00:19:17.905>not <00:19:18.080>the <00:19:18.255>same

351
00:19:20.485 --> 00:19:21.585
<00:19:20.485>as <00:19:20.815>your <00:19:21.145>training <00:19:21.475>sets,

352
00:19:22.885 --> 00:19:26.305
<00:19:22.885>but <00:19:23.085>you <00:19:23.285>also <00:19:23.485>have <00:19:23.685>to <00:19:23.885>be, <00:19:24.085>uh, <00:19:24.285>aware

353
00:19:26.495 --> 00:19:28.745
<00:19:26.495>that <00:19:26.774>you <00:19:27.054>don't <00:19:27.334>go <00:19:27.614>under <00:19:27.895>fit.

354
00:19:28.805 --> 00:19:30.705
<00:19:28.805>So <00:19:29.085>their <00:19:29.365>line <00:19:29.645>is <00:19:29.925>just <00:19:30.205>somewhere

355
00:19:31.095 --> 00:19:34.025
<00:19:31.095>because <00:19:31.295>then <00:19:31.495>it <00:19:31.695>doesn't <00:19:31.895>make <00:19:32.095>any <00:19:32.295>sense <00:19:32.495>anymore.

356
00:19:35.165 --> 00:19:37.985
<00:19:35.165>And <00:19:35.320>the <00:19:35.476>same <00:19:35.631>with <00:19:35.787>regression <00:19:35.942>at <00:19:36.098>the <00:19:36.253>bottom <00:19:36.409>line <00:19:36.565>there.

357
00:19:38.085 --> 00:19:40.225
<00:19:38.085>See <00:19:38.285>the <00:19:38.485>first <00:19:38.685>one <00:19:38.885>with <00:19:39.085>o <00:19:39.285>fittings <00:19:39.485>trying

358
00:19:40.285 --> 00:19:41.865
<00:19:40.285>to <00:19:40.635>hit <00:19:40.985>every <00:19:41.335>data <00:19:41.685>points.

359
00:19:42.645 --> 00:19:43.665
<00:19:42.645>That's <00:19:42.874>not <00:19:43.104>what <00:19:43.333>we <00:19:43.562>want.

360
00:19:43.665 --> 00:19:45.305
<00:19:43.665>We <00:19:43.945>want <00:19:44.225>a <00:19:44.504>smooth <00:19:44.784>little <00:19:45.065>line,

361
00:19:46.265 --> 00:19:49.525
<00:19:46.265>but <00:19:46.420>not <00:19:46.576>a <00:19:46.731>random <00:19:46.887>line <00:19:47.042>that <00:19:47.198>doesn't <00:19:47.353>say <00:19:47.509>anything <00:19:47.665>anymore.

362
00:19:51.665 --> 00:19:55.205
<00:19:51.665>So <00:19:51.865>you <00:19:52.065>can <00:19:52.264>also <00:19:52.465>use <00:19:52.665>the <00:19:52.865>confusion <00:19:53.065>matrix

363
00:19:56.345 --> 00:19:58.085
<00:19:56.345>and <00:19:56.625>maybe <00:19:56.905>you're <00:19:57.185>familiar <00:19:57.465>with <00:19:57.745>it,

364
00:19:58.385 --> 00:20:00.925
<00:19:58.385>but <00:19:58.540>that's <00:19:58.696>more <00:19:58.851>like <00:19:59.007>I <00:19:59.162>predicted <00:19:59.318>that <00:19:59.473>is <00:19:59.629>a <00:19:59.785>cat

365
00:20:01.145 --> 00:20:02.165
<00:20:01.145>and <00:20:01.374>it's <00:20:01.604>actual <00:20:01.833>a <00:20:02.062>cat.

366
00:20:02.465 --> 00:20:03.725
<00:20:02.465>So <00:20:02.748>that's <00:20:03.032>a <00:20:03.315>true <00:20:03.599>positive.

367
00:20:04.185 --> 00:20:08.605
<00:20:04.185>But <00:20:04.360>we <00:20:04.534>can <00:20:04.710>also <00:20:04.885>say <00:20:05.060>it's <00:20:05.235>predicted, <00:20:05.409>uh, <00:20:05.585>dog,

368
00:20:08.785 --> 00:20:10.045
<00:20:08.785>but <00:20:09.068>it's <00:20:09.352>actual <00:20:09.635>a <00:20:09.919>cat.

369
00:20:10.105 --> 00:20:12.765
<00:20:10.105>So <00:20:10.305>that's <00:20:10.505>in <00:20:10.705>false <00:20:10.905>negative <00:20:11.105>and <00:20:11.305>so <00:20:11.505>on.

370
00:20:14.065 --> 00:20:18.845
<00:20:14.065>And <00:20:14.192>based <00:20:14.319>on <00:20:14.446>this <00:20:14.574>you <00:20:14.701>can <00:20:14.828>determine <00:20:14.955>where <00:20:15.083>do <00:20:15.210>wanna <00:20:15.337>focus <00:20:15.465>on.

371
00:20:19.225 --> 00:20:20.685
<00:20:19.225>So <00:20:19.444>if <00:20:19.663>you're <00:20:19.881>on <00:20:20.101>the <00:20:20.320>medical <00:20:20.539>spectrum,

372
00:20:21.215 --> 00:20:25.765
<00:20:21.215>maybe <00:20:21.448>you're <00:20:21.681>happier <00:20:21.915>if <00:20:22.148>someone's <00:20:22.381>diagnosed <00:20:22.615>with

373
00:20:26.305 --> 00:20:28.845
<00:20:26.305>uh, <00:20:26.538>a <00:20:26.771>sickness <00:20:27.005>and <00:20:27.238>doesn't <00:20:27.471>have <00:20:27.705>it.

374
00:20:29.705 --> 00:20:32.245
<00:20:29.705>Or <00:20:29.879>then <00:20:30.054>the <00:20:30.230>other <00:20:30.405>way <00:20:30.580>around <00:20:30.754>it <00:20:30.929>isn't <00:20:31.105>diagnosed

375
00:20:32.245 --> 00:20:35.365
<00:20:32.245>with <00:20:32.524>sickness <00:20:32.804>and <00:20:33.084>it's <00:20:33.364>going, <00:20:33.645>uh,

376
00:20:36.345 --> 00:20:38.845
<00:20:36.345>and <00:20:36.500>it's <00:20:36.656>uh, <00:20:36.811>and <00:20:36.967>the <00:20:37.122>patient <00:20:37.278>is <00:20:37.433>going <00:20:37.589>to <00:20:37.745>die.

377
00:20:38.865 --> 00:20:42.085
<00:20:38.865>But <00:20:38.992>it <00:20:39.119>really <00:20:39.246>depends <00:20:39.374>on <00:20:39.501>the <00:20:39.628>data <00:20:39.755>you <00:20:39.883>have <00:20:40.010>what <00:20:40.137>you <00:20:40.265>want

378
00:20:43.985 --> 00:20:47.845
<00:20:43.985>and <00:20:44.112>there <00:20:44.239>are <00:20:44.366>a <00:20:44.494>lot <00:20:44.621>of <00:20:44.748>formulas <00:20:44.875>you <00:20:45.003>can <00:20:45.130>put <00:20:45.257>onto <00:20:45.385>this.

379
00:20:48.305 --> 00:20:51.605
<00:20:48.305>So <00:20:48.445>the <00:20:48.585>specifically <00:20:48.725>the <00:20:48.865>recall <00:20:49.005>and <00:20:49.145>all <00:20:49.285>that <00:20:49.425>kind <00:20:49.565>of <00:20:49.705>stuff.

380
00:20:52.545 --> 00:20:55.845
<00:20:52.545>And <00:20:52.720>luckily <00:20:52.895>you <00:20:53.070>don't <00:20:53.245>have <00:20:53.420>to <00:20:53.595>remember <00:20:53.770>this <00:20:53.945>also

381
00:20:56.075 --> 00:21:00.885
<00:20:56.075>because <00:20:56.230>there <00:20:56.386>are <00:20:56.541>medals <00:20:56.697>for <00:20:56.852>it <00:20:57.008>And <00:20:57.163>my <00:20:57.319>personal <00:20:57.475>favorite,

382
00:21:01.425 --> 00:21:03.485
<00:21:01.425>the <00:21:01.891>mean <00:21:02.358>squared <00:21:02.825>error

383
00:21:04.545 --> 00:21:06.765
<00:21:04.545>and <00:21:04.778>maybe <00:21:05.011>think, <00:21:05.245>what <00:21:05.478>is <00:21:05.711>this <00:21:05.945>formula?

384
00:21:07.285 --> 00:21:08.885
<00:21:07.285>I <00:21:07.518>will <00:21:07.751>just <00:21:07.985>get <00:21:08.218>it <00:21:08.451>in <00:21:08.685>pieces.

385
00:21:09.945 --> 00:21:13.485
<00:21:09.945>So <00:21:10.061>we <00:21:10.178>say <00:21:10.294>the <00:21:10.411>sum <00:21:10.528>we <00:21:10.645>have <00:21:10.761>ends <00:21:10.878>every <00:21:10.995>data <00:21:11.111>point <00:21:11.228>in <00:21:11.345>our

386
00:21:13.555 --> 00:21:16.085
<00:21:13.555>dataset <00:21:13.905>for <00:21:14.255>Y <00:21:14.605>is <00:21:14.955>one.

387
00:21:16.145 --> 00:21:17.725
<00:21:16.145>So <00:21:16.345>we <00:21:16.545>are <00:21:16.745>going <00:21:16.945>to <00:21:17.145>loop <00:21:17.345>through <00:21:17.545>it

388
00:21:18.825 --> 00:21:21.405
<00:21:18.825>and <00:21:19.058>we're <00:21:19.291>saying, <00:21:19.525>so <00:21:19.758>for <00:21:19.991>every <00:21:20.225>dataset,

389
00:21:21.815 --> 00:21:25.645
<00:21:21.815>every <00:21:21.970>point <00:21:22.126>in <00:21:22.281>the <00:21:22.437>dataset, <00:21:22.592>we <00:21:22.748>cannot <00:21:22.903>calculate <00:21:23.059>the <00:21:23.215>error.

390
00:21:26.105 --> 00:21:29.365
<00:21:26.105>And <00:21:26.245>the <00:21:26.385>error <00:21:26.525>is <00:21:26.665>the <00:21:26.805>difference <00:21:26.945>between <00:21:27.085>Y <00:21:27.225>and <00:21:27.365>y <00:21:27.505>hat.

391
00:21:29.745 --> 00:21:33.765
<00:21:29.745>And <00:21:29.861>y <00:21:29.978>hat <00:21:30.094>is <00:21:30.211>the <00:21:30.328>predicted <00:21:30.445>one <00:21:30.561>and <00:21:30.678>y <00:21:30.794>is <00:21:30.911>the <00:21:31.028>actual <00:21:31.145>one.

392
00:21:34.475 --> 00:21:35.685
<00:21:34.475>Everybody <00:21:35.019>with <00:21:35.563>squares.

393
00:21:36.585 --> 00:21:40.925
<00:21:36.585>So <00:21:36.785>if <00:21:36.985>we <00:21:37.185>see <00:21:37.385>that <00:21:37.585>in <00:21:37.785>pictures <00:21:37.985>here,

394
00:21:40.925 --> 00:21:42.045
<00:21:40.925>you <00:21:41.177>see <00:21:41.429>some <00:21:41.681>data <00:21:41.933>points.

395
00:21:42.715 --> 00:21:47.045
<00:21:42.715>Blue <00:21:42.870>are <00:21:43.026>the <00:21:43.181>uh, <00:21:43.337>are <00:21:43.492>the <00:21:43.648>values, <00:21:43.803>that's <00:21:43.959>the <00:21:44.115>y

396
00:21:47.145 --> 00:21:49.925
<00:21:47.145>and <00:21:47.378>the <00:21:47.611>regression <00:21:47.845>line, <00:21:48.078>that's <00:21:48.311>the <00:21:48.545>predicted.

397
00:21:50.785 --> 00:21:53.725
<00:21:50.785>And <00:21:50.960>then <00:21:51.135>the <00:21:51.310>difference <00:21:51.485>between <00:21:51.660>those <00:21:51.835>are <00:21:52.010>the <00:21:52.185>arrows

398
00:21:53.725 --> 00:21:55.085
<00:21:53.725>and <00:21:54.031>we <00:21:54.337>can <00:21:54.643>calculate <00:21:54.949>that

399
00:21:55.085 --> 00:21:57.485
<00:21:55.085>for <00:21:55.285>every <00:21:55.485>data <00:21:55.685>points <00:21:55.885>in <00:21:56.085>our <00:21:56.285>data <00:21:56.485>sets.

400
00:21:57.985 --> 00:21:59.685
<00:21:57.985>And <00:21:58.185>then <00:21:58.385>we <00:21:58.584>get <00:21:58.784>the <00:21:58.985>mean <00:21:59.185>square <00:21:59.385>error.

401
00:22:01.425 --> 00:22:04.685
<00:22:01.425>So <00:22:01.552>if <00:22:01.679>you <00:22:01.806>do <00:22:01.934>that <00:22:02.061>in <00:22:02.188>code, <00:22:02.315>it <00:22:02.443>looks <00:22:02.570>something <00:22:02.697>like <00:22:02.825>this,

402
00:22:04.785 --> 00:22:07.165
<00:22:04.785>we <00:22:05.018>can <00:22:05.251>just <00:22:05.485>call <00:22:05.718>the <00:22:05.951>method <00:22:06.185>luckily

403
00:22:08.425 --> 00:22:10.685
<00:22:08.425>and <00:22:08.891>uh, <00:22:09.358>print <00:22:09.825>it

404
00:22:11.185 --> 00:22:14.605
<00:22:11.185>and <00:22:11.385>then <00:22:11.585>that's <00:22:11.784>print <00:22:11.985>something <00:22:12.185>like <00:22:12.385>this, <00:22:12.585>0.56

405
00:22:14.985 --> 00:22:16.925
<00:22:14.985>and <00:22:15.334>then <00:22:15.685>you're, <00:22:16.034>okay, <00:22:16.385>cool.

406
00:22:17.485 --> 00:22:18.765
<00:22:17.485>I <00:22:17.629>have <00:22:17.773>no <00:22:17.917>clue <00:22:18.061>what <00:22:18.205>to <00:22:18.349>do <00:22:18.493>with <00:22:18.637>this.

407
00:22:20.385 --> 00:22:23.645
<00:22:20.385>So <00:22:20.618>in <00:22:20.851>this <00:22:21.085>example, <00:22:21.318>I <00:22:21.551>used, <00:22:21.785>uh,

408
00:22:24.105 --> 00:22:25.765
<00:22:24.105>the <00:22:24.571>California <00:22:25.038>housing <00:22:25.505>set

409
00:22:26.705 --> 00:22:29.725
<00:22:26.705>and <00:22:26.860>here <00:22:27.016>we <00:22:27.171>wanna <00:22:27.327>predict <00:22:27.482>the <00:22:27.638>prices <00:22:27.793>of <00:22:27.949>the <00:22:28.105>houses

410
00:22:30.905 --> 00:22:33.885
<00:22:30.905>and <00:22:31.080>the <00:22:31.254>prices <00:22:31.430>were <00:22:31.605>in <00:22:31.780>hundred <00:22:31.955>thousands <00:22:32.129>of <00:22:32.305>dollars.

411
00:22:34.345 --> 00:22:38.805
<00:22:34.345>So <00:22:34.485>if <00:22:34.625>we <00:22:34.765>do <00:22:34.905>hundred <00:22:35.045>thousands <00:22:35.185>of <00:22:35.325>dollars <00:22:35.465>times <00:22:35.605>the <00:22:35.745>0.56,

412
00:22:39.475 --> 00:22:43.565
<00:22:39.475>then <00:22:39.708>we <00:22:39.941>got <00:22:40.175>65,000 <00:22:40.408>error <00:22:40.641>on <00:22:40.875>efforts.

413
00:22:44.145 --> 00:22:45.605
<00:22:44.145>And <00:22:44.407>I <00:22:44.670>know <00:22:44.933>you <00:22:45.196>know <00:22:45.459>you,

414
00:22:45.905 --> 00:22:48.605
<00:22:45.905>but <00:22:46.060>I <00:22:46.216>think <00:22:46.371>that's <00:22:46.527>a <00:22:46.682>big <00:22:46.838>difference <00:22:46.993>in <00:22:47.149>housing <00:22:47.305>prices.

415
00:22:49.145 --> 00:22:51.845
<00:22:49.145>So <00:22:49.285>here <00:22:49.425>we're <00:22:49.565>thinking, <00:22:49.705>okay, <00:22:49.845>we <00:22:49.985>need <00:22:50.125>to <00:22:50.264>do <00:22:50.405>something <00:22:50.545>else

416
00:22:51.945 --> 00:22:53.445
<00:22:51.945>and <00:22:52.215>improve <00:22:52.485>this <00:22:52.754>model <00:22:53.024>because <00:22:53.294>it's

417
00:22:53.825 --> 00:22:54.965
<00:22:53.825>way <00:22:54.081>too <00:22:54.338>big <00:22:54.594>of <00:22:54.851>difference.

418
00:22:58.335 --> 00:23:02.475
<00:22:58.335>Um, <00:22:58.475>yeah, <00:22:58.615>the <00:22:58.755>accuracy, <00:22:58.895>you <00:22:59.035>can <00:22:59.175>see <00:22:59.315>here <00:22:59.455>an <00:22:59.595>example, <00:22:59.735>uh,

419
00:23:03.265 --> 00:23:04.275
<00:23:03.265>also <00:23:03.568>how <00:23:03.871>it's <00:23:04.174>doing.

420
00:23:06.255 --> 00:23:09.035
<00:23:06.255>And <00:23:06.430>we <00:23:06.605>can <00:23:06.780>do <00:23:06.955>the <00:23:07.130>confu <00:23:07.305>matrix <00:23:07.480>as <00:23:07.655>well.

421
00:23:10.135 --> 00:23:14.435
<00:23:10.135>So <00:23:10.275>there's <00:23:10.415>a <00:23:10.555>lot <00:23:10.695>of <00:23:10.835>things. <00:23:10.975>So <00:23:11.115>now <00:23:11.254>we <00:23:11.395>see, <00:23:11.535>okay,

422
00:23:14.455 --> 00:23:17.275
<00:23:14.455>the <00:23:14.629>model <00:23:14.804>has <00:23:14.980>to <00:23:15.155>be <00:23:15.330>better, <00:23:15.504>let's <00:23:15.679>improve <00:23:15.855>it.

423
00:23:18.695 --> 00:23:20.475
<00:23:18.695>So <00:23:19.044>for <00:23:19.395>improving <00:23:19.745>the <00:23:20.095>model,

424
00:23:20.895 --> 00:23:23.075
<00:23:20.895>you <00:23:21.095>can <00:23:21.295>do <00:23:21.495>a <00:23:21.695>lot <00:23:21.895>of <00:23:22.095>different <00:23:22.295>things.

425
00:23:23.775 --> 00:23:26.235
<00:23:23.775>We <00:23:23.975>can <00:23:24.175>choose <00:23:24.375>another <00:23:24.575>model, <00:23:24.775>collect <00:23:24.975>more <00:23:25.175>data,

426
00:23:26.295 --> 00:23:27.755
<00:23:26.295>that's <00:23:26.623>always <00:23:26.952>a <00:23:27.280>good <00:23:27.609>idea.

427
00:23:28.775 --> 00:23:32.955
<00:23:28.775>Um, <00:23:28.930>handling <00:23:29.086>the <00:23:29.241>outliers, <00:23:29.397>lots <00:23:29.552>of <00:23:29.708>things <00:23:29.863>you <00:23:30.019>can <00:23:30.175>do.

428
00:23:34.175 --> 00:23:37.115
<00:23:34.175>So <00:23:34.330>for <00:23:34.486>example, <00:23:34.641>if <00:23:34.797>you're <00:23:34.952>taking <00:23:35.108>the <00:23:35.263>ling, <00:23:35.419>the <00:23:35.575>outliers,

429
00:23:38.025 --> 00:23:42.035
<00:23:38.025>then <00:23:38.141>we <00:23:38.258>say <00:23:38.375>we <00:23:38.491>use <00:23:38.608>wind, <00:23:38.725>ice, <00:23:38.841>that's <00:23:38.958>a <00:23:39.075>method <00:23:39.191>to <00:23:39.308>do <00:23:39.425>that.

430
00:23:42.655 --> 00:23:46.045
<00:23:42.655>And <00:23:42.782>then <00:23:42.909>we <00:23:43.036>can <00:23:43.164>say, <00:23:43.291>okay, <00:23:43.418>for <00:23:43.545>the <00:23:43.673>price <00:23:43.800>it's <00:23:43.927>limit <00:23:44.055>it

431
00:23:46.145 --> 00:23:49.325
<00:23:46.145>and <00:23:46.320>take <00:23:46.495>the <00:23:46.670>top <00:23:46.845>and <00:23:47.020>the <00:23:47.195>bottom <00:23:47.370>5% <00:23:47.545>away.

432
00:23:49.425 --> 00:23:51.285
<00:23:49.425>So <00:23:49.658>not <00:23:49.891>the, <00:23:50.125>the <00:23:50.358>bridges <00:23:50.591>of <00:23:50.825>the,

433
00:23:51.285 --> 00:23:53.685
<00:23:51.285>the <00:23:51.460>boxes <00:23:51.635>on <00:23:51.810>the <00:23:51.985>bridges <00:23:52.160>and <00:23:52.335>not <00:23:52.510>the <00:23:52.685>fillas.

434
00:23:53.985 --> 00:23:56.645
<00:23:53.985>That's <00:23:54.159>not <00:23:54.334>very, <00:23:54.510>uh, <00:23:54.685>logical <00:23:54.860>for <00:23:55.034>the <00:23:55.209>house <00:23:55.385>prices.

435
00:23:57.385 --> 00:24:01.325
<00:23:57.385>So <00:23:57.585>please <00:23:57.785>drop <00:23:57.985>them. <00:23:58.185>And <00:23:58.385>luckily <00:23:58.585>you <00:23:58.785>can

436
00:24:01.425 --> 00:24:02.645
<00:24:01.425>do <00:24:01.791>this <00:24:02.157>at <00:24:02.523>home.

437
00:24:02.825 --> 00:24:05.485
<00:24:02.825>So <00:24:02.952>if <00:24:03.079>you <00:24:03.206>scan <00:24:03.334>this <00:24:03.461>qr, <00:24:03.588>it'll <00:24:03.715>also <00:24:03.843>be <00:24:03.970>at <00:24:04.097>the <00:24:04.225>end

438
00:24:05.485 --> 00:24:08.525
<00:24:05.485>of <00:24:05.685>the <00:24:05.885>presentation, <00:24:06.084>you <00:24:06.284>go <00:24:06.485>to <00:24:06.685>my <00:24:06.885>GitHub

439
00:24:08.865 --> 00:24:11.245
<00:24:08.865>and <00:24:09.040>then <00:24:09.215>you <00:24:09.390>can <00:24:09.565>uh, <00:24:09.740>try <00:24:09.915>to <00:24:10.090>with <00:24:10.265>play

440
00:24:11.245 --> 00:24:12.685
<00:24:11.245>with <00:24:11.569>it <00:24:11.893>such <00:24:12.216>Jupyter <00:24:12.541>notebook.

441
00:24:12.945 --> 00:24:15.125
<00:24:12.945>So <00:24:13.145>you <00:24:13.345>only <00:24:13.544>need <00:24:13.745>or <00:24:13.945>a <00:24:14.145>Google <00:24:14.345>account

442
00:24:15.905 --> 00:24:17.845
<00:24:15.905>or <00:24:16.138>uh, <00:24:16.371>admin <00:24:16.605>rights <00:24:16.838>on <00:24:17.071>your <00:24:17.305>laptop.

443
00:24:17.985 --> 00:24:19.205
<00:24:17.985>You're <00:24:18.351>trying <00:24:18.716>this <00:24:19.082>on?

444
00:24:26.075 --> 00:24:28.175
<00:24:26.075>Yes. <00:24:26.308>Okay. <00:24:26.541>It's <00:24:26.775>also <00:24:27.008>on <00:24:27.241>the <00:24:27.475>end.

445
00:24:28.195 --> 00:24:29.615
<00:24:28.195>So <00:24:28.377>if <00:24:28.560>you <00:24:28.742>didn't <00:24:28.925>get <00:24:29.107>it, <00:24:29.290>no <00:24:29.473>worries.

446
00:24:31.195 --> 00:24:34.575
<00:24:31.195>Um, <00:24:31.370>real <00:24:31.544>life <00:24:31.720>examples, <00:24:31.895>you <00:24:32.070>probably <00:24:32.245>know <00:24:32.419>them <00:24:32.595>like,

447
00:24:34.715 --> 00:24:35.895
<00:24:34.715>uh, <00:24:35.069>Amazon <00:24:35.423>and <00:24:35.777>Netflix.

448
00:24:36.005 --> 00:24:37.095
<00:24:36.005>They <00:24:36.201>collect <00:24:36.397>a <00:24:36.593>lot <00:24:36.789>of <00:24:36.986>data

449
00:24:37.235 --> 00:24:38.495
<00:24:37.235>and <00:24:37.518>then <00:24:37.802>they <00:24:38.085>get, <00:24:38.369>uh,

450
00:24:38.805 --> 00:24:41.375
<00:24:38.805>predictions <00:24:39.085>like <00:24:39.365>maybe <00:24:39.645>also <00:24:39.925>like <00:24:40.205>this.

451
00:24:42.555 --> 00:24:45.815
<00:24:42.555>And <00:24:42.682>here <00:24:42.809>on <00:24:42.936>the <00:24:43.064>left <00:24:43.191>hand <00:24:43.318>side, <00:24:43.445>you <00:24:43.573>see <00:24:43.700>your <00:24:43.827>right <00:24:43.955>hands

452
00:24:46.915 --> 00:24:50.335
<00:24:46.915>and <00:24:47.115>sorry, <00:24:47.315>um, <00:24:47.514>you <00:24:47.715>see <00:24:47.915>the <00:24:48.115>prediction <00:24:48.315>of

453
00:24:50.355 --> 00:24:52.615
<00:24:50.355>how <00:24:50.635>good <00:24:50.915>the <00:24:51.195>bone <00:24:51.475>structure <00:24:51.755>is.

454
00:24:55.675 --> 00:25:00.335
<00:24:55.675>So <00:24:55.802>I <00:24:55.929>hope <00:24:56.056>in <00:24:56.184>this <00:24:56.311>30 <00:24:56.438>minutes <00:24:56.565>you <00:24:56.693>learn, <00:24:56.820>first <00:24:56.947>of <00:24:57.075>all,

455
00:25:00.485 --> 00:25:04.615
<00:25:00.485>collect <00:25:00.659>a <00:25:00.834>lot <00:25:01.010>of <00:25:01.185>data, <00:25:01.360>don't <00:25:01.534>invent <00:25:01.709>everything <00:25:01.885>yourself.

456
00:25:04.615 --> 00:25:06.135
<00:25:04.615>There's <00:25:04.888>a <00:25:05.162>lot <00:25:05.435>already <00:25:05.709>out <00:25:05.983>there.

457
00:25:07.325 --> 00:25:09.535
<00:25:07.325>Machine <00:25:07.605>learning <00:25:07.885>isn't <00:25:08.165>always <00:25:08.445>the <00:25:08.725>way

458
00:25:10.435 --> 00:25:12.015
<00:25:10.435>and <00:25:10.784>please <00:25:11.135>verify <00:25:11.485>the <00:25:11.835>model,

459
00:25:12.075 --> 00:25:14.975
<00:25:12.075>but <00:25:12.230>also <00:25:12.386>in <00:25:12.541>the <00:25:12.697>verifying <00:25:12.852>it's <00:25:13.008>gonna <00:25:13.163>take <00:25:13.319>a <00:25:13.475>lot

460
00:25:15.435 --> 00:25:19.935
<00:25:15.435>and <00:25:15.562>a <00:25:15.689>lot <00:25:15.816>of <00:25:15.944>time <00:25:16.071>and <00:25:16.198>money <00:25:16.325>to, <00:25:16.453>uh, <00:25:16.580>to <00:25:16.707>do <00:25:16.835>that.

461
00:25:21.475 --> 00:25:24.895
<00:25:21.475>So <00:25:21.708>the <00:25:21.941>wrap <00:25:22.175>up, <00:25:22.408>you're <00:25:22.641>probably <00:25:22.875>wondering

462
00:25:24.895 --> 00:25:26.335
<00:25:24.895>what <00:25:25.154>happened <00:25:25.413>to <00:25:25.672>the <00:25:25.931>original <00:25:26.191>problem.

463
00:25:26.435 --> 00:25:28.095
<00:25:26.435>We <00:25:26.668>want <00:25:26.901>to <00:25:27.135>see <00:25:27.368>how <00:25:27.601>it <00:25:27.835>works.

464
00:25:28.915 --> 00:25:31.855
<00:25:28.915>So <00:25:29.042>due <00:25:29.169>to <00:25:29.296>the <00:25:29.424>reasons, <00:25:29.551>I <00:25:29.678>don't <00:25:29.805>have <00:25:29.933>those <00:25:30.060>two <00:25:30.187>cats <00:25:30.315>anymore,

465
00:25:33.235 --> 00:25:37.325
<00:25:33.235>but <00:25:34.635>I'm

466
00:25:37.325 --> 00:25:38.405
<00:25:37.325>walking <00:25:37.649>too <00:25:37.973>far <00:25:38.297>away.

467
00:25:39.245 --> 00:25:41.245
<00:25:39.245>I <00:25:39.445>have <00:25:39.645>new <00:25:39.844>cats <00:25:40.044>and <00:25:40.245>she <00:25:40.445>has <00:25:40.645>kittens.

468
00:25:41.825 --> 00:25:43.165
<00:25:41.825>So <00:25:42.126>there's <00:25:42.428>three <00:25:42.729>weeks <00:25:43.031>old.

469
00:25:43.705 --> 00:25:45.525
<00:25:43.705>So <00:25:43.938>if <00:25:44.171>they <00:25:44.405>don't <00:25:44.638>drink <00:25:44.871>milk <00:25:45.105>anymore,

470
00:25:45.725 --> 00:25:47.005
<00:25:45.725>I <00:25:45.955>have <00:25:46.185>the <00:25:46.416>same <00:25:46.646>problem <00:25:46.877>again.

471
00:25:47.505 --> 00:25:50.605
<00:25:47.505>So <00:25:47.660>next <00:25:47.816>time <00:25:47.971>I <00:25:48.127>will <00:25:48.282>tell <00:25:48.438>you <00:25:48.593>all <00:25:48.749>about <00:25:48.905>it

472
00:25:50.825 --> 00:25:54.805
<00:25:50.825>and <00:25:50.952>see <00:25:51.079>you, <00:25:51.206>uh, <00:25:51.334>what <00:25:51.461>the <00:25:51.588>solution <00:25:51.715>was <00:25:51.843>to <00:25:51.970>my <00:25:52.097>hobby <00:25:52.225>project.

473
00:25:57.105 --> 00:26:00.525
<00:25:57.105>Um, <00:25:57.305>Feedback. <00:25:57.505>I <00:25:57.705>always <00:25:57.905>love <00:25:58.105>to <00:25:58.305>improve <00:25:58.505>myself.

474
00:26:00.585 --> 00:26:04.125
<00:26:00.585>So <00:26:00.740>on <00:26:00.896>the <00:26:01.051>left <00:26:01.207>hand <00:26:01.362>you <00:26:01.518>have <00:26:01.673>cure <00:26:01.829>for <00:26:01.985>feedback

475
00:26:04.705 --> 00:26:07.285
<00:26:04.705>and <00:26:04.845>on <00:26:04.985>the <00:26:05.125>right <00:26:05.264>hand <00:26:05.405>the <00:26:05.544>hands <00:26:05.685>on <00:26:05.824>coat <00:26:05.965>from <00:26:06.105>before.

476
00:26:08.385 --> 00:26:10.525
<00:26:08.385>Um, <00:26:08.665>and <00:26:08.945>my <00:26:09.225>kitten <00:26:09.504>that <00:26:09.785>really,

477
00:26:10.665 --> 00:26:12.605
<00:26:10.665>and <00:26:10.865>I <00:26:11.065>love <00:26:11.264>if <00:26:11.465>you <00:26:11.665>fill <00:26:11.865>that <00:26:12.065>in.

478
00:26:14.995 --> 00:26:16.095
<00:26:14.995>So <00:26:15.242>thank <00:26:15.490>you <00:26:15.737>very <00:26:15.985>much.

479
00:26:16.795 --> 00:26:21.255
<00:26:16.795>I'm, <00:26:16.911>and <00:26:17.028>I'll <00:26:17.145>be <00:26:17.261>here <00:26:17.378>all <00:26:17.495>day <00:26:17.611>and <00:26:17.728>tomorrow <00:26:17.845>if <00:26:17.961>you <00:26:18.078>have <00:26:18.195>q

480
00:26:21.255 --> 00:26:21.935
<00:26:21.255>and <00:26:21.561>a <00:26:21.867>questions.