﻿1
00:00:00,433 --> 00:00:02,933
Welcome back to your Machine Learning
A-To-Z course.

2
00:00:02,933 --> 00:00:04,900
In this tutorial,
I'd like to give you a quick heads

3
00:00:04,900 --> 00:00:08,233
up about the non-linear support
vector regression.

4
00:00:08,800 --> 00:00:12,400
So what we discussed
previously was intuition for SVR.

5
00:00:12,400 --> 00:00:12,666
Indeed.

6
00:00:12,666 --> 00:00:17,400
But this was intuition of the linear
support vector regression model.

7
00:00:17,733 --> 00:00:22,566
And while it's a great start into
the world of SVR, it is just the start.

8
00:00:22,566 --> 00:00:25,266
There is so much more to support
vector regression.

9
00:00:25,266 --> 00:00:28,166
this was a simple linear model.

10
00:00:28,166 --> 00:00:30,366
There are actually other models,

11
00:00:30,366 --> 00:00:34,533
and the rest assured
they will be covered in this course.

12
00:00:34,633 --> 00:00:38,633
So they are coming up
and we'll be talking about,

13
00:00:38,633 --> 00:00:43,266
for instance, the RBF kernel, SVR,
and the kernel trick

14
00:00:43,266 --> 00:00:47,866
and all these important things
that make this a model even more robust.

15
00:00:48,200 --> 00:00:51,500
However, I wanted to give you
this heads up because already in

16
00:00:51,500 --> 00:00:55,066
upcoming Python tutorials
that are coming right after.

17
00:00:55,300 --> 00:00:57,866
Yeah, just following this tutorial,
in this section,

18
00:00:57,866 --> 00:01:01,033
you'll be working with
had learned to create your first,

19
00:01:01,033 --> 00:01:05,900
SVR model, and the SVR model
will already be nonlinear.

20
00:01:05,900 --> 00:01:07,400
So here you can see the results.

21
00:01:07,400 --> 00:01:09,266
Like I've blurred
them out on purpose because,

22
00:01:10,800 --> 00:01:11,766
I wanted not to.

23
00:01:11,766 --> 00:01:13,200
I wanted to make sure you don't get,

24
00:01:13,200 --> 00:01:16,733
a spoiler by seeing the results
in terms of the numbers,

25
00:01:17,100 --> 00:01:21,666
but you can see the model itself here,
and you can see that it's non-linear.

26
00:01:21,666 --> 00:01:23,266
So just a straight line.

27
00:01:23,266 --> 00:01:26,700
And that is because we're using
or at least will be showing you

28
00:01:26,700 --> 00:01:32,400
how to use, the radial basis
function, kernel for this SVR.

29
00:01:32,400 --> 00:01:34,933
So it's really going to be
a nonlinear SVR.

30
00:01:34,933 --> 00:01:37,133
So now what's the reason for this?

31
00:01:37,133 --> 00:01:37,500
heads up.

32
00:01:37,500 --> 00:01:43,300
Well, the thing is that the non linear
support vector, regression

33
00:01:43,300 --> 00:01:47,333
will be kind of further down the course
beyond this section.

34
00:01:47,633 --> 00:01:50,400
So it's a bit of a mismatch,
but that's, done on purpose

35
00:01:50,400 --> 00:01:53,400
because that's the easier way
to grasp these concepts.

36
00:01:53,666 --> 00:01:55,466
And I'll explain in this following slides.

37
00:01:55,466 --> 00:01:56,900
So this is what

38
00:01:58,400 --> 00:01:58,666
this is

39
00:01:58,666 --> 00:02:01,666
what the nonlinear SVR will look like.

40
00:02:01,666 --> 00:02:01,833
Right.

41
00:02:01,833 --> 00:02:02,900
So compared to this

42
00:02:02,900 --> 00:02:06,400
this is your linear model as you can see
is much more complex over here.

43
00:02:06,400 --> 00:02:09,100
We have to go into a third dimension
and perform

44
00:02:09,100 --> 00:02:11,700
some calculations
or some modeling there and go back.

45
00:02:11,700 --> 00:02:13,133
So it is quite involved.

46
00:02:13,133 --> 00:02:17,666
And in order to get there, 
the following tutorials are essential.

47
00:02:17,666 --> 00:02:21,733
So first in the section on SVM you'll find
a tutorial called SVM intuition.

48
00:02:21,866 --> 00:02:25,800
So that's an essential tutorial
to take in the steps

49
00:02:25,800 --> 00:02:31,200
to getting to kernel SVR, 
or nonlinear SVR.

50
00:02:31,300 --> 00:02:35,100
Then in this section on kernel SVM,
the following section, you

51
00:02:35,100 --> 00:02:38,166
can find them in the part of the course
that is to do with classification.

52
00:02:38,400 --> 00:02:42,766
So in the section on kernel SVM there's
a tool called kernel SVM intuition.

53
00:02:42,766 --> 00:02:44,900
Then there's mapping
to a higher dimension.

54
00:02:44,900 --> 00:02:47,333
The kernel trick types of kernel function.

55
00:02:47,333 --> 00:02:50,566
And then finally nonlinear kernel SVR.

56
00:02:50,566 --> 00:02:53,400
So as you can see it's quite a complex
topics topic.

57
00:02:53,400 --> 00:02:55,733
And it comes at the very end of this list.

58
00:02:55,733 --> 00:02:57,933
So from here you have two choices.

59
00:02:57,933 --> 00:03:02,133
You can either just after this tutorial
go straight to advanced Python

60
00:03:02,133 --> 00:03:06,566
tutorials on kernel on the SVR there
that are coming up and just do them.

61
00:03:06,733 --> 00:03:07,800
And just keep in mind

62
00:03:07,800 --> 00:03:12,200
that he's using a non-linear kernel,
and that will come later in the course.

63
00:03:12,200 --> 00:03:13,666
The intuition for it will come later.

64
00:03:13,666 --> 00:03:14,333
And of course,

65
00:03:14,333 --> 00:03:17,566
or if you would like to get the intuition
before you do

66
00:03:17,566 --> 00:03:18,800
the practical side of things,

67
00:03:18,800 --> 00:03:22,966
then it's essential that you would need
to go through all of these tutorials

68
00:03:22,966 --> 00:03:24,933
that are mentioned on this slide over
here.

69
00:03:24,933 --> 00:03:27,966
Get to non-linear kernel SVR
and then come back to the Python terms.

70
00:03:28,200 --> 00:03:29,333
So there are two options.

71
00:03:29,333 --> 00:03:33,900
the my preference if you ask me I would
just go for advanced Python tutorials.

72
00:03:33,900 --> 00:03:37,733
Now just keep in mind that he's using
a non-linear kernel along with the code,

73
00:03:37,733 --> 00:03:41,466
and then you will still get the intuition
when the time is right when you get to

74
00:03:41,466 --> 00:03:43,333
this section on kernel SVM.

75
00:03:43,333 --> 00:03:43,866
So there you go.

76
00:03:43,866 --> 00:03:46,066
That's just a quick heads up on
what's coming up.

77
00:03:46,066 --> 00:03:47,433
of course the choice is yours.

78
00:03:47,433 --> 00:03:50,333
And in any case, please enjoy the course.

79
00:03:50,333 --> 00:03:52,900
And there's lots of fun
tutorials coming up,

80
00:03:52,900 --> 00:03:54,300
and I look forward to seeing you next one.

81
00:03:54,300 --> 00:03:56,033
Until then, happy analyzing.
