1
00:00:00,540 --> 00:00:01,620
Welcome back.

2
00:00:01,620 --> 00:00:06,500
In the last video we saw a few different ways that we can create an umpire race.

3
00:00:06,570 --> 00:00:09,840
So we got a sample array which could really just be a simple array.

4
00:00:09,870 --> 00:00:15,660
But we're using NDP dot array passing and a list of numbers as another one if we wanted to create an

5
00:00:15,660 --> 00:00:22,000
empty array of just ones we can pass it a shape and if we keep going if we want an array of zeros we

6
00:00:22,000 --> 00:00:24,160
can use empty dot zeros.

7
00:00:24,200 --> 00:00:30,100
We wanted an array that's between a range of different numbers say starts at zero stops at ten steps

8
00:00:30,110 --> 00:00:36,020
at two we can use a range and if we wanted an array of random numbers we've got a few options.

9
00:00:36,070 --> 00:00:43,270
We have Rand int for returning integers between a certain range or we have random at random for returning

10
00:00:43,300 --> 00:00:50,140
numbers between 0 1 of a certain shape and how do we know what these functions do well we compress shift

11
00:00:50,140 --> 00:00:50,740
and tab.

12
00:00:50,740 --> 00:00:52,660
And that's going to return the doc string.

13
00:00:52,660 --> 00:01:00,220
So here we can see N.P. dot random dot random return random floats in the half open interval 0 to 1

14
00:01:00,670 --> 00:01:06,400
and it takes a parameter called size which in our case we've done five and three so five rows three

15
00:01:06,400 --> 00:01:10,380
columns we can see that they're wonderful.

16
00:01:10,420 --> 00:01:17,550
Now he also said in the previous video that none pi generating random numbers they aren't actually random

17
00:01:17,620 --> 00:01:19,500
but you might be running this cell.

18
00:01:19,510 --> 00:01:24,830
I've run this one about 50 odd times and each time new random numbers have come up.

19
00:01:24,880 --> 00:01:26,770
So what's the deal what gives Daniel why.

20
00:01:26,770 --> 00:01:28,720
I think these numbers are random.

21
00:01:28,720 --> 00:01:30,060
Well you'd be right.

22
00:01:30,070 --> 00:01:37,030
They look random to us but what they really are is pseudo random numbers and we can see this by using

23
00:01:37,060 --> 00:01:40,420
num pi random dot seed.

24
00:01:40,420 --> 00:01:45,100
Now throughout your machine learning endeavors you might see this little function at the start of a

25
00:01:45,100 --> 00:01:51,340
notebook or somewhere along lines wherever num plays random package is being used.

26
00:01:51,640 --> 00:01:53,860
Now what does this do now rather than talk about it.

27
00:01:53,860 --> 00:01:56,610
We're going to see what it does in a practical sense.

28
00:01:56,800 --> 00:02:02,280
So let's just comment this out for the time being and we'll just create another one.

29
00:02:02,330 --> 00:02:08,060
You go random array for just so we can make sure it's not the same as this one as before and paid at

30
00:02:08,060 --> 00:02:19,100
random the rand let's go Rand int we want numbers up to 10 of size 5 three wonderful and then we'll

31
00:02:19,100 --> 00:02:23,070
have a look at what this looks like random away for every guy.

32
00:02:23,200 --> 00:02:24,190
Okay numbers.

33
00:02:24,190 --> 00:02:29,120
These are between 0 and 10 0 5 beautiful of size.

34
00:02:29,150 --> 00:02:30,160
Well this is really shape.

35
00:02:30,160 --> 00:02:31,840
That's a bit confusing with implies

36
00:02:34,730 --> 00:02:42,010
is that the size parameter of this creating array function is actually the shape attribute.

37
00:02:42,020 --> 00:02:43,150
That's right.

38
00:02:43,280 --> 00:02:47,810
We can figure these things out with moment to do that but every time if we ran this we're gonna get

39
00:02:48,910 --> 00:02:56,020
what seems to be a new random array and for what we're looking at we can consider this random numbers

40
00:02:56,020 --> 00:02:57,000
every time.

41
00:02:57,190 --> 00:03:04,370
But what if we wanted to share our notebook with someone else and we wanted them to run our cells across

42
00:03:04,370 --> 00:03:09,960
the notebook go through from top to bottom and get exactly the same results as what we did.

43
00:03:10,000 --> 00:03:12,640
So we can reproduce our experiments.

44
00:03:12,790 --> 00:03:20,320
That's when num PIs random seed comes in it allows us to still generate random numbers but random numbers

45
00:03:20,320 --> 00:03:22,180
that are reproducible.

46
00:03:22,180 --> 00:03:23,550
So let's have a look.

47
00:03:23,710 --> 00:03:31,090
If we set random seed at the top of this cell to be zero so we can go shift tab takes a parameter seed

48
00:03:31,120 --> 00:03:33,070
equals something.

49
00:03:33,070 --> 00:03:36,180
So we can actually set this to any number we want.

50
00:03:36,220 --> 00:03:37,360
We'll have to look at any second.

51
00:03:37,360 --> 00:03:39,660
So in this case we'll set it to zero.

52
00:03:39,880 --> 00:03:49,210
We'll hit shift and enter what's happened five out on the numbers of they've stopped changing on us

53
00:03:50,560 --> 00:03:51,480
now what if we comment.

54
00:03:51,480 --> 00:03:54,270
Nissan they've changed.

55
00:03:54,990 --> 00:04:01,540
Okay so what the random seed function does is it essentially says hey um hi.

56
00:04:01,680 --> 00:04:07,140
I want you to create random numbers but keep them aligned with the seed.

57
00:04:07,260 --> 00:04:09,690
So random seed is zero.

58
00:04:09,690 --> 00:04:15,150
We're always gonna see this combination 5 0 3 5 0 3 5 0 3.

59
00:04:15,240 --> 00:04:16,210
I'm running this cell.

60
00:04:16,230 --> 00:04:18,340
We can see the numbers going up here now.

61
00:04:18,690 --> 00:04:24,970
This works no matter what you'll see it is so 6 3 7 6 3 7.

62
00:04:24,990 --> 00:04:32,440
We can't even set it to now and I know 9 0 3 1 0 3 1 huh.

63
00:04:32,790 --> 00:04:36,180
Now let's try again with another random function that was reined in.

64
00:04:36,690 --> 00:04:38,180
Maybe we go here.

65
00:04:38,460 --> 00:04:45,300
Let's go random array 5 just to make sure that it's different NPC not random dot random and shift tab.

66
00:04:45,300 --> 00:04:47,730
What does this take size equals something.

67
00:04:47,790 --> 00:04:49,610
So we want 5 and 3 again.

68
00:04:49,680 --> 00:04:50,700
We had a view.

69
00:04:50,770 --> 00:04:52,160
What's going on here.

70
00:04:52,200 --> 00:04:53,100
Random array.

71
00:04:53,100 --> 00:04:53,760
5.

72
00:04:54,750 --> 00:04:55,050
Okay.

73
00:04:55,070 --> 00:04:57,650
Every time we run this new numbers.

74
00:04:58,320 --> 00:04:59,370
Okay.

75
00:04:59,670 --> 00:05:02,580
Now let's put a random seed in here.

76
00:05:02,700 --> 00:05:03,750
What should we do.

77
00:05:03,750 --> 00:05:05,400
What number are you thinking of.

78
00:05:05,400 --> 00:05:06,900
I'm thinking of seven.

79
00:05:06,930 --> 00:05:10,410
There we got seven.

80
00:05:10,740 --> 00:05:17,720
The numbers still seem seemingly random to me but what the random seed does is remember it says an umpire.

81
00:05:17,730 --> 00:05:23,010
Hey generate some random numbers but if you see the random scene it set to 7.

82
00:05:23,040 --> 00:05:27,830
Make sure they're saying every time the random seed is set to 7.

83
00:05:27,870 --> 00:05:42,550
So now if we run this exact same code in the next cell random array 5 equals a random random 5 3.

84
00:05:42,640 --> 00:05:44,110
Now we want to see it again.

85
00:05:44,120 --> 00:05:47,810
N.P. array 5 different here.

86
00:05:50,950 --> 00:05:57,130
If we kept going with this we would eventually line up with the random seed but we don't have time to

87
00:05:57,130 --> 00:05:57,520
do that.

88
00:05:58,760 --> 00:06:00,470
So what's the takeaway here.

89
00:06:00,500 --> 00:06:09,150
Well the numbers that you see num pi generate seemingly random but they're not actually random.

90
00:06:09,200 --> 00:06:11,290
They're pseudo random numbers.

91
00:06:11,390 --> 00:06:18,170
So whenever we run this function without num pi seed being run what it's actually doing is setting the

92
00:06:18,170 --> 00:06:21,920
seed to be something random every single time you run this.

93
00:06:22,040 --> 00:06:27,540
But if we set our seed manually we're going to get the same random numbers back.

94
00:06:27,740 --> 00:06:35,120
Now the important takeaway is that these numbers are seemingly random but the main purpose of using

95
00:06:35,120 --> 00:06:41,530
random seed is if we wanted someone else to be able to use our notebook and generate random numbers.

96
00:06:41,630 --> 00:06:48,830
But the same random numbers as US and thus reproduce our experiment exactly we would set a random seed.

97
00:06:48,830 --> 00:06:51,980
So if you see any random seed that's what it's doing.

98
00:06:51,980 --> 00:06:58,730
It's making sure the random numbers in the cell that it's been run are the same every single time.

99
00:06:59,400 --> 00:07:02,540
OK now we've seen that we've seen how to create some arrays.

100
00:07:02,540 --> 00:07:05,250
We've seen how to generate arrays with random numbers.

101
00:07:05,660 --> 00:07:09,660
Let's look at how we can view different arrays and matrices.

102
00:07:09,740 --> 00:07:10,970
So take a little break.

103
00:07:10,970 --> 00:07:16,130
Try to generate some more random arrays practice around with no high random seed and I'll see you in

104
00:07:16,130 --> 00:07:16,850
the next video.