﻿1
00:00:00,200 --> 00:00:00,966
All right.

2
00:00:00,966 --> 00:00:02,633
So now all good.

3
00:00:02,633 --> 00:00:06,800
Now we apply correctly this transformation
to one hot

4
00:00:06,800 --> 00:00:10,166
encode
the columns of this matrix of features x.

5
00:00:10,533 --> 00:00:14,266
Let's check that right away
by creating a new code cell

6
00:00:14,266 --> 00:00:17,933
and then printing this
matrix of features x.

7
00:00:18,300 --> 00:00:19,266
There we go.

8
00:00:19,266 --> 00:00:21,900
Now let's run this cell.

9
00:00:21,900 --> 00:00:24,900
And then running the next one.

10
00:00:24,900 --> 00:00:26,333
And what do we see.

11
00:00:26,333 --> 00:00:29,333
Well we see exactly
what was expected here.

12
00:00:29,500 --> 00:00:33,300
Meaning that we no longer have
the first country column with the text.

13
00:00:33,300 --> 00:00:35,800
You know, the three countries here
and strings.

14
00:00:35,800 --> 00:00:38,666
This time we have,
as we said, three new columns.

15
00:00:38,666 --> 00:00:40,800
So I don't know if you can see it,
but that's the first one.

16
00:00:40,800 --> 00:00:42,066
That's the second one.

17
00:00:42,066 --> 00:00:46,700
And that's the third one where each
row encodes one of the three countries.

18
00:00:46,900 --> 00:00:49,300
So actually France, you know,
that was the first row.

19
00:00:49,300 --> 00:00:53,300
France is encoded as one zero and zero.

20
00:00:53,300 --> 00:00:55,366
You know a vector of one zero and zero.

21
00:00:55,366 --> 00:00:59,566
Then Spain is encoded
as a vector of zero zero and one.

22
00:00:59,866 --> 00:01:03,733
And Germany is encoded as a vector of zero
one and zero.

23
00:01:03,766 --> 00:01:07,633
You see the unique idea
for each of these three countries.

24
00:01:07,633 --> 00:01:09,333
Thanks to these three columns here.

25
00:01:09,333 --> 00:01:11,966
And that's exactly
the idea of one hot encoding.

26
00:01:11,966 --> 00:01:14,933
We not only turn our countries

27
00:01:14,933 --> 00:01:19,033
into numerical values,
but also there is not a numerical order.

28
00:01:19,266 --> 00:01:22,000
Thanks to these zeros
and ones here in three columns.

29
00:01:22,000 --> 00:01:23,000
So that's perfect.

30
00:01:23,000 --> 00:01:26,933
And that will provide the best results
for our future machine learning models.

31
00:01:27,633 --> 00:01:29,500
All right. So congratulations.

32
00:01:29,500 --> 00:01:32,966
Now you know how to one hot encode
some categorical data.

33
00:01:33,500 --> 00:01:35,933
Now we're going to quickly
do another encoding

34
00:01:35,933 --> 00:01:39,800
for the dependent variable
because indeed it has a text format.

35
00:01:39,833 --> 00:01:41,066
No. And yes.

36
00:01:41,066 --> 00:01:45,700
And we would just like to convert these
strings into zero and one respectively.

37
00:01:45,933 --> 00:01:47,700
And to do this well that's very simple.

38
00:01:47,700 --> 00:01:51,566
We're going to use another class
called label encoder and which will

39
00:01:51,566 --> 00:01:56,800
exactly encode these nos and yeses
into zeros and ones respectively.

40
00:01:57,233 --> 00:01:57,533
All right.

41
00:01:57,533 --> 00:01:58,200
So let's do this.

42
00:01:58,200 --> 00:02:00,766
Let's create a new code cell here.

43
00:02:00,766 --> 00:02:03,033
Let's scroll down a bit.

44
00:02:03,033 --> 00:02:04,066
Here we go.

45
00:02:04,066 --> 00:02:07,000
And now let's encode
the dependent variable.

46
00:02:07,000 --> 00:02:10,800
So as we said we're going to use the label
encoder class which we get

47
00:02:10,966 --> 00:02:15,633
from the scikit learn library again
from which we're going to call

48
00:02:16,033 --> 00:02:21,166
the preprocessing module
from which we're going to import.

49
00:02:22,500 --> 00:02:26,833
Well there we go the label encoder class.

50
00:02:27,366 --> 00:02:30,600
Then exactly as what we did
before, we're going to create

51
00:02:30,600 --> 00:02:33,600
an object of this class
which we're going to call elite.

52
00:02:33,766 --> 00:02:38,500
And to do this well we simply need to call
the class label encoder.

53
00:02:38,900 --> 00:02:41,766
There we go then some parentheses.

54
00:02:41,766 --> 00:02:42,600
Then good news.

55
00:02:42,600 --> 00:02:44,733
We don't have to input anything
in the parentheses

56
00:02:44,733 --> 00:02:48,533
because you know
we will just need to enter directly. Why.

57
00:02:48,566 --> 00:02:50,733
Because it is only one single vector.

58
00:02:50,733 --> 00:02:53,733
So it will be obvious
what will be needed to encode.

59
00:02:53,866 --> 00:02:55,833
And so there we go. Let's do this.

60
00:02:55,833 --> 00:03:00,300
Let's first call our object early
from which we're going to call.

61
00:03:00,300 --> 00:03:01,133
Well good news.

62
00:03:01,133 --> 00:03:04,266
Once again there is a fit transform method

63
00:03:04,600 --> 00:03:07,966
which we can call directly on Y

64
00:03:08,366 --> 00:03:11,500
and which will exactly convert the nose.

65
00:03:11,500 --> 00:03:14,500
And yes,
you know the text inside numerical values.

66
00:03:14,800 --> 00:03:17,366
And this time
we don't have to have a numpy array

67
00:03:17,366 --> 00:03:19,433
because this is the dependent
variable vector.

68
00:03:19,433 --> 00:03:21,366
It doesn't need to be a numpy array.

69
00:03:21,366 --> 00:03:24,533
You know as what is expected
by the future machinery models.

70
00:03:24,900 --> 00:03:30,533
So we can just set the new Y to be
what is returned by this fit

71
00:03:30,566 --> 00:03:34,266
transform method applied to the old way
with the Texas nose.

72
00:03:34,266 --> 00:03:35,200
And yes.

73
00:03:35,200 --> 00:03:36,400
All right, let's check this.

74
00:03:36,400 --> 00:03:39,900
Let's create a new code cell just to print

75
00:03:40,500 --> 00:03:42,900
our dependent variable vector y.

76
00:03:42,900 --> 00:03:47,233
Let's run this cell first
and then this one.

77
00:03:47,233 --> 00:03:48,800
And let's see if we get zeros and ones.

78
00:03:48,800 --> 00:03:51,800
And yes, indeed
we get the zeros and the ones.

79
00:03:51,900 --> 00:03:54,400
Right. No. Corresponds to zero.

80
00:03:54,400 --> 00:03:56,833
And then one here
correspond to. Yes. Right.

81
00:03:56,833 --> 00:04:02,233
Then it should be 0011000110.

82
00:04:02,266 --> 00:04:03,666
Okay. So all good.

83
00:04:03,666 --> 00:04:07,033
So now you not only know how to apply
one hot encoding

84
00:04:07,033 --> 00:04:08,666
when you have several categories.

85
00:04:08,666 --> 00:04:11,200
And one of the features
of your matrix of features.

86
00:04:11,200 --> 00:04:13,000
But also you can do a simple label

87
00:04:13,000 --> 00:04:16,966
encoding when you have, you know,
two classes which you can directly encode

88
00:04:16,966 --> 00:04:19,966
into zeros and ones,
you know, a binary outcome.

89
00:04:20,700 --> 00:04:22,766
All right. Perfect. Congratulations.

90
00:04:22,766 --> 00:04:26,266
Now you have an extra tool in your data
preprocessing toolkit.

91
00:04:26,400 --> 00:04:28,566
The way to encode categorical data.

92
00:04:28,566 --> 00:04:32,633
And now we're going to move on
to the next tool which will be to split

93
00:04:32,633 --> 00:04:35,633
the data
set into the training set and test set.

94
00:04:35,633 --> 00:04:36,700
So digest this.

95
00:04:36,700 --> 00:04:39,533
And as soon as you're ready
meet me in the next tutorial.
