1
00:00:00,133 --> 00:00:00,966
Hello my friends.

2
00:00:00,966 --> 00:00:02,233
Happy to see you again.

3
00:00:02,233 --> 00:00:06,166
Now for this new section
and mostly this new practical activity

4
00:00:06,300 --> 00:00:10,166
where we're going to build together
the random forest regression.

5
00:00:10,500 --> 00:00:13,500
So this is our last model
of the regression part.

6
00:00:13,500 --> 00:00:16,633
And we're going to build it
very quickly and efficiently.

7
00:00:16,633 --> 00:00:21,266
Because actually it looks very much like
the decision tree regression model.

8
00:00:21,400 --> 00:00:22,933
So we will be very efficient.

9
00:00:22,933 --> 00:00:27,366
And besides right after this section comes
the most important section for you.

10
00:00:27,633 --> 00:00:33,000
It is the section where Kirill and I will
explain to you how to not only evaluate

11
00:00:33,000 --> 00:00:36,533
your regression models,
but also how to select the best one.

12
00:00:36,766 --> 00:00:39,733
So right after we're done with the random
forest regression,

13
00:00:39,733 --> 00:00:43,700
we will actually have a new data set with,
you know, multiple features.

14
00:00:43,700 --> 00:00:46,000
You know, like more real world data set.

15
00:00:46,000 --> 00:00:51,866
And I will show you how to use your
regression code templates to quickly plug

16
00:00:51,900 --> 00:00:56,266
your regression models onto the data set
and quickly find the best one.

17
00:00:56,300 --> 00:00:56,666
All right.

18
00:00:56,666 --> 00:00:59,966
So that will be very important
for you to learn how to handle.

19
00:01:00,100 --> 00:01:04,833
But now let's smash our last regression
model the random forest regression.

20
00:01:05,133 --> 00:01:08,633
And before we start, let's just make sure
everyone here is on the same page.

21
00:01:08,833 --> 00:01:13,066
I give you the link to this folder
right before this tutorial in the article.

22
00:01:13,100 --> 00:01:14,966
So make sure to connect to that link.

23
00:01:14,966 --> 00:01:16,866
And now we should all be on the same page.

24
00:01:16,866 --> 00:01:19,866
So we're all going to go into part
two regression.

25
00:01:20,100 --> 00:01:20,900
Then we're going to go

26
00:01:20,900 --> 00:01:24,866
this time to the last regression model
which is our random forest regression.

27
00:01:25,100 --> 00:01:28,366
And then we're going to start
as usual with Python.

28
00:01:28,900 --> 00:01:32,566
And this folder contains once again
the position salaries data set.

29
00:01:32,766 --> 00:01:37,833
And of course the random forest regression
implementation in the Ipynb format,

30
00:01:37,833 --> 00:01:42,400
which you can therefore open with either
Google Colaboratory or Jupyter Notebook.

31
00:01:42,600 --> 00:01:46,033
And as far as I'm concerned, as usual
I'm going to open it with

32
00:01:46,266 --> 00:01:47,966
Google Colaboratory.

33
00:01:47,966 --> 00:01:48,966
So there we go.

34
00:01:48,966 --> 00:01:53,400
Let's start implementing
the random forest regression model.

35
00:01:53,700 --> 00:01:57,533
So now it is,
you know, laying out the notebook.

36
00:01:58,066 --> 00:01:59,800
And in a second we should have it.

37
00:01:59,800 --> 00:02:00,966
There we go.

38
00:02:00,966 --> 00:02:01,600
All right.

39
00:02:01,600 --> 00:02:04,533
So as usual this is in read only mode.

40
00:02:04,533 --> 00:02:09,166
So we're going to quickly create a copy
so that we can re-implement it.

41
00:02:09,400 --> 00:02:13,400
We're not going to re-implement it
from scratch because it is really similar

42
00:02:13,533 --> 00:02:16,333
as the decision tree regression model.

43
00:02:16,333 --> 00:02:20,200
So you will see that
we will only re-implement one code cell.

44
00:02:20,200 --> 00:02:22,833
And if you know, you come here
in this regression

45
00:02:22,833 --> 00:02:25,966
folder for the first time
with random forest regression, well,

46
00:02:25,966 --> 00:02:29,700
I encourage you to have a look at decision
tree regression first, because indeed

47
00:02:30,000 --> 00:02:32,000
all these code cells were explained right.

48
00:02:32,000 --> 00:02:35,433
But this time
we're only going to delete this one.

49
00:02:35,733 --> 00:02:36,533
That's where we,

50
00:02:36,533 --> 00:02:40,366
you know, train the random forest
regression model on the whole data set.

51
00:02:40,600 --> 00:02:42,300
And then all the rest is same.

52
00:02:42,300 --> 00:02:45,366
It is all the same as decision tree
regression.

53
00:02:45,366 --> 00:02:48,600
We import first the libraries,
then we import the data set.

54
00:02:49,000 --> 00:02:51,700
Then after, you know, the training
of the random forest regression

55
00:02:51,700 --> 00:02:53,266
model on the whole data set,

56
00:02:53,266 --> 00:02:57,333
we predict this new result,
which is the exact same syntax.

57
00:02:57,333 --> 00:02:59,800
Actually I'm going to hide this now.

58
00:02:59,800 --> 00:03:03,666
So this is the exact same syntax
as the decision tree regression model.

59
00:03:03,766 --> 00:03:05,400
And then here that's the same.

60
00:03:05,400 --> 00:03:08,533
This is the exact same code
that we implement

61
00:03:08,533 --> 00:03:11,533
to visualize the random forest
regression result.

62
00:03:11,566 --> 00:03:11,900
All right.

63
00:03:11,900 --> 00:03:14,666
So let's just keep this
because we did it many times.

64
00:03:14,666 --> 00:03:18,600
And I'm sure you're looking forward
to that final section

65
00:03:18,600 --> 00:03:21,000
where you know
everything is going to make sense.

66
00:03:21,000 --> 00:03:24,633
Because indeed
you will learn how to use this regression

67
00:03:24,633 --> 00:03:27,833
folder containing
all these code templates for regression.

68
00:03:27,833 --> 00:03:30,966
And you will learn how to understand
which model to choose.

69
00:03:30,966 --> 00:03:33,666
And, you know, select the best
one for your data set.

70
00:03:33,666 --> 00:03:36,366
I will explain everything
in this last section,

71
00:03:36,366 --> 00:03:40,333
but for now, let's implement
that only missing code

72
00:03:40,333 --> 00:03:44,500
cell to train the random forest
regression model on the whole dataset.

73
00:03:44,800 --> 00:03:46,900
So let's add a new code cell here.

74
00:03:46,900 --> 00:03:47,700
And now there you go.

75
00:03:47,700 --> 00:03:48,733
You could once again

76
00:03:48,733 --> 00:03:53,033
totally do it yourself
by looking at some documentation online.

77
00:03:53,033 --> 00:03:54,933
And actually, well, let's do it together.

78
00:03:54,933 --> 00:03:58,800
This time we're going to pretend that,
you know, we would like to build a random

79
00:03:58,800 --> 00:04:01,333
forest regression model
and train it on the data set,

80
00:04:01,333 --> 00:04:05,200
and that I have absolutely no clue on
how to build it, or you know,

81
00:04:05,266 --> 00:04:08,300
which Scikit-Learn class
to use to build it.

82
00:04:08,633 --> 00:04:09,433
So let's see.

83
00:04:09,433 --> 00:04:14,166
Well, what I would do, as I said,
would be to go to Google or Bing.

84
00:04:14,633 --> 00:04:15,400
So here is Google.

85
00:04:15,400 --> 00:04:17,566
And I would type here in the search bar.

86
00:04:17,566 --> 00:04:20,566
Well you know for example scikit learn.

87
00:04:20,900 --> 00:04:21,400
All right.

88
00:04:21,400 --> 00:04:25,300
And then random forest regression
this one right.

89
00:04:25,666 --> 00:04:28,033
Even the suggestion helps.
So that's perfect.

90
00:04:28,033 --> 00:04:29,366
Then pressing enter.

91
00:04:29,366 --> 00:04:32,800
And then I would go for the link of scikit
learn.

92
00:04:32,800 --> 00:04:34,266
You know the scikit learn website.

93
00:04:34,266 --> 00:04:37,666
And this will usually be the first link
as it is the case right here.

94
00:04:37,666 --> 00:04:39,233
So let's click this.

95
00:04:39,233 --> 00:04:44,700
And normally I should find side the exact
name of the random forest regression

96
00:04:44,700 --> 00:04:48,633
class, and also the name of the module
that contains this class.

97
00:04:48,633 --> 00:04:52,100
And indeed
that is exactly what we see in big here.

98
00:04:52,333 --> 00:04:54,366
This is the whole library scikit learn.

99
00:04:54,366 --> 00:04:56,966
This is the module that contains
the class we want.

100
00:04:56,966 --> 00:04:59,200
And this is the name of the class.

101
00:04:59,200 --> 00:05:00,366
So there you go.

102
00:05:00,366 --> 00:05:03,300
Well let's actually take everything here.

103
00:05:03,300 --> 00:05:04,800
And we will get this bit.

104
00:05:04,800 --> 00:05:07,800
Because this is not exactly
what we have to write in Python.

105
00:05:07,800 --> 00:05:10,533
But I'm sure you'll know
how to adapt this.

106
00:05:10,533 --> 00:05:11,100
There you go.

107
00:05:11,100 --> 00:05:12,266
That's what I wanted to show you.

108
00:05:12,266 --> 00:05:15,600
You know, it's very,
very easy to find online.

109
00:05:15,600 --> 00:05:19,200
The name of class that allows to build
the model you want, right?

110
00:05:19,200 --> 00:05:22,266
I just had to type scikit
learn and random forest regression

111
00:05:22,500 --> 00:05:24,000
and just go to the first link.

112
00:05:24,000 --> 00:05:26,000
So you see very very easy.

113
00:05:26,000 --> 00:05:29,300
So let's go back to our implementation
and let's start building

114
00:05:29,433 --> 00:05:33,433
this random forest regression model
and train it on the whole data set.

115
00:05:33,700 --> 00:05:35,833
So now I'm going to paste
what I've just copied.

116
00:05:35,833 --> 00:05:39,333
Because indeed,
in order to import this class

117
00:05:39,333 --> 00:05:44,133
I simply need to add here at the beginning
of from right from scikit learn

118
00:05:44,500 --> 00:05:47,633
and then from the
in symbol module of scikit learn.

119
00:05:48,000 --> 00:05:52,833
Well I'm going to import
that random forest regressor class.