1
00:00:00,133 --> 00:00:02,433
Hello and welcome to this art tutorial.

2
00:00:02,433 --> 00:00:05,933
So today we are going to implement
our second nonlinear regression model.

3
00:00:06,000 --> 00:00:09,233
It's going to be the SVR support
vector for regression.

4
00:00:09,533 --> 00:00:10,433
So let's do it.

5
00:00:10,433 --> 00:00:12,366
We're going to use a regression template.

6
00:00:12,366 --> 00:00:15,366
And you're going to see
how it's going to be so easy.

7
00:00:15,566 --> 00:00:17,100
So let's start by doing the basics.

8
00:00:17,100 --> 00:00:18,966
Set the right folder as working directory.

9
00:00:18,966 --> 00:00:21,966
So we'll go to Machine
learning A-Z part two regression.

10
00:00:22,366 --> 00:00:25,400
And then support vector regression right.

11
00:00:25,433 --> 00:00:26,466
That's the right folder.

12
00:00:26,466 --> 00:00:27,400
All good.

13
00:00:27,400 --> 00:00:31,600
Let's click on set as working directory
to set the folder as a working directory.

14
00:00:32,066 --> 00:00:34,000
Great. And now we can start.

15
00:00:34,000 --> 00:00:36,366
So we're going to take our regression
templates.

16
00:00:36,366 --> 00:00:39,933
And we're going to take everything
from here to here.

17
00:00:40,500 --> 00:00:43,666
Copy and paste it in our SVR model.

18
00:00:44,266 --> 00:00:45,066
Here we go.

19
00:00:45,066 --> 00:00:47,133
And now we just need
to change a few things.

20
00:00:47,133 --> 00:00:50,133
So let's start by changing the basics.

21
00:00:50,266 --> 00:00:53,266
We're going to replace
regression model here by SVR.

22
00:00:53,766 --> 00:00:54,200
Okay.

23
00:00:55,166 --> 00:00:59,400
And same here SVR results.

24
00:01:00,100 --> 00:01:02,900
Okay. And same here.

25
00:01:02,900 --> 00:01:04,233
All right.

26
00:01:04,233 --> 00:01:05,466
That was the easy step.

27
00:01:05,466 --> 00:01:07,800
And now let's get to the interesting step.

28
00:01:07,800 --> 00:01:11,633
So this interesting step is
of course to create our SVR regressor.

29
00:01:11,900 --> 00:01:13,333
And we are creating it here.

30
00:01:13,333 --> 00:01:15,433
So I'm going to remove this.

31
00:01:15,433 --> 00:01:18,100
And now let's create this regressor.

32
00:01:18,100 --> 00:01:20,433
It's going to take 3 or 4 lines as usual.

33
00:01:20,433 --> 00:01:21,466
It's going to be very simple.

34
00:01:21,466 --> 00:01:25,200
We're going to use a function
which is the SVM function,

35
00:01:25,200 --> 00:01:29,400
very simply because the SVR you know
is a support vector machine algorithm.

36
00:01:29,700 --> 00:01:32,700
But for regression,
that's why we call it SVR.

37
00:01:32,700 --> 00:01:35,700
And therefore we're just taking it
from the SVM function.

38
00:01:35,700 --> 00:01:38,400
And you will understand why perfectly.

39
00:01:38,400 --> 00:01:42,600
So first let's import the required package
because this function

40
00:01:42,600 --> 00:01:46,133
is contained in a package
which is the E10 71 package.

41
00:01:46,166 --> 00:01:50,400
So let's go to our package is here
and let's check if we have it.

42
00:01:51,266 --> 00:01:54,833
So I have it because of course I used it
before, but you might not have it.

43
00:01:54,833 --> 00:01:59,100
So in case you don't have it, I'm
going to type this line here

44
00:01:59,266 --> 00:02:03,300
that you will need to execute
to install this package.

45
00:02:03,566 --> 00:02:06,133
So you type install that package is here.

46
00:02:06,133 --> 00:02:09,166
And then in the parenthesis
and in the quotes you type

47
00:02:09,166 --> 00:02:12,533
the name of the package which is E10 71.

48
00:02:13,066 --> 00:02:13,500
All right.

49
00:02:13,500 --> 00:02:16,500
And then you just select
this line and execute.

50
00:02:16,666 --> 00:02:18,266
And this will install the package.

51
00:02:18,266 --> 00:02:20,900
I'm not going to do it
because mine is already installed.

52
00:02:20,900 --> 00:02:23,900
And therefore I'm
going to put that in comment.

53
00:02:23,933 --> 00:02:26,766
Just press command plus shift plus C.

54
00:02:26,766 --> 00:02:27,666
All right.

55
00:02:27,666 --> 00:02:31,333
And now let's actually
also add this line library.

56
00:02:32,333 --> 00:02:36,066
And in parenthesis E10 71 not in quotes.

57
00:02:36,900 --> 00:02:40,233
And this will select
automatically the package here

58
00:02:40,466 --> 00:02:43,966
E10 71
because this will not always be selected.

59
00:02:44,266 --> 00:02:46,966
But by including that in your script
you will be all fine.

60
00:02:46,966 --> 00:02:49,433
This will always be selected. Okay.

61
00:02:49,433 --> 00:02:53,100
And now let's start
creating our SVR regressor.

62
00:02:53,400 --> 00:02:56,733
So as usual we're going to start
by defining our regressor

63
00:02:57,600 --> 00:03:00,600
regressor and then equals.

64
00:03:00,733 --> 00:03:03,866
And then as I mentioned before
we're going to use the SVM function

65
00:03:04,100 --> 00:03:06,300
which is actually this.

66
00:03:06,300 --> 00:03:07,266
And then in parentheses.

67
00:03:07,266 --> 00:03:09,466
And now let's press here

68
00:03:09,466 --> 00:03:13,466
F1 to have a look at the arguments
and see what we need to input.

69
00:03:14,066 --> 00:03:16,100
So the first argument is formula.

70
00:03:16,100 --> 00:03:18,033
So you start to know perfectly what it is.

71
00:03:18,033 --> 00:03:21,033
Of course the formula is

72
00:03:21,733 --> 00:03:23,333
our dependent variable

73
00:03:23,333 --> 00:03:26,100
which is remember salary

74
00:03:26,100 --> 00:03:30,600
and then tilde I just press alt n
and then a dot here.

75
00:03:30,600 --> 00:03:34,066
And the dot specifies that we are taking
all the independent variable

76
00:03:34,400 --> 00:03:35,633
of our data set.

77
00:03:36,766 --> 00:03:37,500
We actually have

78
00:03:37,500 --> 00:03:40,533
one independent variable
which is the level independent variable.

79
00:03:40,566 --> 00:03:42,600
So we could also type here level.

80
00:03:42,600 --> 00:03:44,966
But most of the time we use the dot
because of course

81
00:03:44,966 --> 00:03:47,966
we have a lot more than one
independent variable.

82
00:03:48,266 --> 00:03:50,933
But as a reminder, I'm
taking one independent variable here

83
00:03:50,933 --> 00:03:53,933
so that we can clearly see
the visual graphic results

84
00:03:54,166 --> 00:03:57,166
of the different nonlinear models
that we are building.

85
00:03:57,466 --> 00:04:00,333
Okay.
So now let's add the second argument.

86
00:04:00,333 --> 00:04:01,800
So I'm adding a comma here.

87
00:04:01,800 --> 00:04:03,533
And then next argument.

88
00:04:03,533 --> 00:04:06,533
So the next argument is data okay

89
00:04:07,000 --> 00:04:10,000
I guess
you know what this data argument will be.

90
00:04:10,500 --> 00:04:14,900
It's going to be
of course our data set data set here

91
00:04:14,900 --> 00:04:18,300
because we did not create any training set
or test set.

92
00:04:18,700 --> 00:04:20,900
You know before we used the training set.

93
00:04:20,900 --> 00:04:23,433
Here is the data. But here
we don't have any training set.

94
00:04:23,433 --> 00:04:26,200
So of course
we're going to take the whole data set

95
00:04:26,200 --> 00:04:29,200
because we want to make very accurate
predictions okay.

96
00:04:29,400 --> 00:04:32,100
And now finally most important argument.

97
00:04:32,100 --> 00:04:33,866
Well all the arguments are important.

98
00:04:33,866 --> 00:04:37,400
But the most relevant for this section
about SVR

99
00:04:37,566 --> 00:04:40,433
is actually the next arguments
we are about to input.

100
00:04:40,433 --> 00:04:43,400
So this argument is actually not x or y.

101
00:04:43,400 --> 00:04:45,400
These are optional arguments.

102
00:04:45,400 --> 00:04:49,900
The most important argument
I was just talking about is this one

103
00:04:49,900 --> 00:04:53,966
the type, because this argument
type will actually specify

104
00:04:53,966 --> 00:04:57,900
if you're making an SVM model,
which is used for classification,

105
00:04:58,233 --> 00:05:01,533
or an SVR model
which is used for regression.

106
00:05:01,900 --> 00:05:04,933
So here, since we're building
a non-linearity regression model,

107
00:05:05,100 --> 00:05:08,266
we will choose the EPS regression type.

108
00:05:08,933 --> 00:05:11,733
We could choose actually EPS regression
or new regression.

109
00:05:11,733 --> 00:05:13,700
But as you can see
we want to take the most common one.

110
00:05:13,700 --> 00:05:15,266
The EPS regression.

111
00:05:15,266 --> 00:05:18,100
And if we were making a SVM model

112
00:05:18,100 --> 00:05:21,933
for classification then we would have
chosen C classification here.

113
00:05:21,933 --> 00:05:25,966
As you can see, association
is the default type for classification.

114
00:05:26,466 --> 00:05:31,566
So in short if you want to do regression
you choose type equals EPS regression.

115
00:05:31,966 --> 00:05:35,766
And if you want to do classification then
you choose type equals C classification.

116
00:05:35,766 --> 00:05:40,100
And actually in the next part about
classification we will make an SVM model.

117
00:05:40,133 --> 00:05:43,800
And for this specific model
we will choose the classification type

118
00:05:44,800 --> 00:05:45,166
okay.

119
00:05:45,166 --> 00:05:46,500
But here we are doing regression.

120
00:05:46,500 --> 00:05:48,500
So we choose the EPS regression type.

121
00:05:48,500 --> 00:05:51,500
So let's input it
I'm going to type here type

122
00:05:51,933 --> 00:05:55,400
equals eps regression.

123
00:05:55,966 --> 00:05:56,366
All right.

124
00:05:56,366 --> 00:05:59,366
And you need to put that in.