﻿1
00:00:00,466 --> 00:00:01,066
And this little

2
00:00:01,066 --> 00:00:04,500
something that we're going to add
is actually the polynomial features.

3
00:00:05,000 --> 00:00:09,000
The polynomial features are actually
some additional independent variables

4
00:00:09,266 --> 00:00:12,366
that are going to be the level squared,
the level cubed level

5
00:00:12,366 --> 00:00:15,800
to the fourth level
to the fifth up to any degree you want.

6
00:00:16,200 --> 00:00:18,833
These are these additional independent
variables.

7
00:00:18,833 --> 00:00:22,466
There are going to compose
our new matrix of features in some way,

8
00:00:22,666 --> 00:00:24,533
which will actually be the matrix

9
00:00:24,533 --> 00:00:28,033
on which we will apply a multiple linear
regression model,

10
00:00:28,366 --> 00:00:31,366
which will make the whole model
a polynomial regression model.

11
00:00:31,666 --> 00:00:36,233
So in short, a polynomial regression
model is a multiple linear regression

12
00:00:36,233 --> 00:00:40,500
model that is composed of one independent
variable and additional independent

13
00:00:40,500 --> 00:00:44,633
variables that are the polynomial terms
of this first independent variable.

14
00:00:45,166 --> 00:00:49,066
Okay, so now that we get this idea,
you'll perfectly understand

15
00:00:49,066 --> 00:00:52,833
what we're about to do right now when
building our polynomial regression model.

16
00:00:53,100 --> 00:00:57,600
Because what we are going to do
is to build this polynomial terms.

17
00:00:57,966 --> 00:01:02,400
And to that extent, what we're
simply going to do is to add specifically

18
00:01:02,633 --> 00:01:06,366
a new column here in this data set,
which will be the level squared.

19
00:01:06,366 --> 00:01:09,866
So we're going to call
this new independent variable level two.

20
00:01:10,266 --> 00:01:12,733
And let's add this new column right now.

21
00:01:12,733 --> 00:01:16,200
So to add a column in a data frame
we need to take our data frame.

22
00:01:16,200 --> 00:01:18,300
So its data set.

23
00:01:18,300 --> 00:01:19,300
Here it is.

24
00:01:19,300 --> 00:01:21,766
And then we need to add
a dollar sign here.

25
00:01:21,766 --> 00:01:26,100
And then we can add any name here which
will create a new column of this name.

26
00:01:26,100 --> 00:01:28,200
And add it into our data set.

27
00:01:28,200 --> 00:01:31,133
So we're going to call this column
level two

28
00:01:31,133 --> 00:01:34,200
because we're taking the level
squared of our ten levels.

29
00:01:34,833 --> 00:01:36,166
Then we add equals.

30
00:01:36,166 --> 00:01:38,466
And then we actually need
to give the formula

31
00:01:38,466 --> 00:01:41,900
of the values in this level
two column of our data set.

32
00:01:42,300 --> 00:01:47,366
And this formula is therefore the levels
in our data set at the power of two.

33
00:01:47,600 --> 00:01:49,666
So very simply we're going to take

34
00:01:49,666 --> 00:01:53,000
all the levels of our data
set by taking our data set.

35
00:01:53,133 --> 00:01:55,200
And same adding this dollar here

36
00:01:55,200 --> 00:01:59,000
to take the level column, which is here,
I just have to press enter.

37
00:01:59,400 --> 00:02:01,900
So by doing this I'm
taking the level column.

38
00:02:01,900 --> 00:02:04,900
That is all the values of the level column
in our data set.

39
00:02:05,266 --> 00:02:09,766
And now I'm going to take the square
of all these levels in our data set.

40
00:02:10,166 --> 00:02:13,166
And so to do this very simply
I just add hat.

41
00:02:13,166 --> 00:02:17,666
And two that's it
that we'll just create a new column

42
00:02:17,666 --> 00:02:20,666
which will contain the squares
of our ten levels in our data set.

43
00:02:20,866 --> 00:02:24,500
So let's see right now
my data set contains only two columns

44
00:02:24,833 --> 00:02:28,200
level and salary. And if I execute this

45
00:02:29,966 --> 00:02:30,766
here we go.

46
00:02:30,766 --> 00:02:32,733
Now let's have a look at the data set.

47
00:02:32,733 --> 00:02:37,533
As you can see I now have three columns
level salary and level two.

48
00:02:37,833 --> 00:02:41,033
And as you can
notice the values in the level two column

49
00:02:41,033 --> 00:02:44,033
are the squares of the values
in the level column.

50
00:02:44,500 --> 00:02:45,833
Okay.

51
00:02:45,833 --> 00:02:48,533
And now we can build
our polynomial regression model.

52
00:02:48,533 --> 00:02:52,833
Because now this dot here
not only contains the level column

53
00:02:53,066 --> 00:02:55,166
but also the level two column.

54
00:02:55,166 --> 00:02:58,266
So this will build a multiple linear
regression model

55
00:02:58,466 --> 00:03:02,900
where the independent variables
are the original independent variable

56
00:03:03,200 --> 00:03:06,733
and the polynomial term
of this first independent variable.

57
00:03:07,000 --> 00:03:09,766
And if you want to build
a polynomial regression

58
00:03:09,766 --> 00:03:12,900
model with a higher degree, well,
you will need to do the same here.

59
00:03:12,900 --> 00:03:16,633
You know,
we can copy this line and paste it here.

60
00:03:16,800 --> 00:03:21,300
And just add a level
three column that will contain the cubes

61
00:03:21,300 --> 00:03:26,100
of the levels in the original independent
variable level of our data set.

62
00:03:26,466 --> 00:03:28,333
And as you can see, it'll be very easy.

63
00:03:28,333 --> 00:03:30,700
I just need to execute this line.

64
00:03:30,700 --> 00:03:34,600
And this will add a level
three column in our data set.

65
00:03:34,600 --> 00:03:37,600
And so now this little dot here will be

66
00:03:37,966 --> 00:03:42,133
the original independent variable level,
the square values of our levels

67
00:03:42,133 --> 00:03:46,600
in level two and the cube values
of our levels in the column level three.

68
00:03:47,400 --> 00:03:51,433
You can add as many powers of levels
as you want, but maybe we will stop here.

69
00:03:51,700 --> 00:03:54,900
We will see what is the best
polynomial regression for our model.

70
00:03:55,066 --> 00:03:57,166
And so we will see what
we'll get with this one.

71
00:03:57,166 --> 00:04:00,800
And now let's actually build
the polynomial regression model.

72
00:04:00,800 --> 00:04:04,833
If we select this and execute press
Command and Control plus enter to execute

73
00:04:05,333 --> 00:04:08,866
and run our polynomial regression
model is created.

74
00:04:09,233 --> 00:04:09,700
Awesome.

75
00:04:09,700 --> 00:04:11,600
So we're going to have a look.

76
00:04:11,600 --> 00:04:14,700
Let's type
summary in the console poly rag.

77
00:04:15,700 --> 00:04:16,966
And let's press enter.

78
00:04:16,966 --> 00:04:18,800
And let's see where we get okay.

79
00:04:18,800 --> 00:04:23,000
So now what we can see is that these level
two and level three polynomial terms

80
00:04:23,000 --> 00:04:26,633
that we created are actually
statistically significant variables.

81
00:04:27,033 --> 00:04:29,866
But actually this table here does not show

82
00:04:29,866 --> 00:04:33,266
the real deal of the polynomial regression
model for our problem.

83
00:04:33,566 --> 00:04:35,133
You will be much more convinced

84
00:04:35,133 --> 00:04:38,700
in the next tutorial, we will be
visualizing the graphic results,

85
00:04:38,933 --> 00:04:41,966
and you will perfectly understand
why our polynomial regression,

86
00:04:41,966 --> 00:04:46,866
which is a non-linear model, will make a
much better job at predicting what we want

87
00:04:47,000 --> 00:04:50,833
compared to this linear regression model,
because it's a linear model.

88
00:04:51,200 --> 00:04:53,266
So we'll check that in the next tutorial.

89
00:04:53,266 --> 00:04:56,266
And until then, enjoy machine learning.
