﻿1
00:00:00,033 --> 00:00:00,966
Hello, my friends.

2
00:00:00,966 --> 00:00:02,100
All right, let's do this.

3
00:00:02,100 --> 00:00:06,133
Let's build and train the decision tree
regression model on the whole data set.

4
00:00:06,366 --> 00:00:09,266
So I actually could ask you to do it
yourself.

5
00:00:09,266 --> 00:00:12,866
You know, even without giving you
the hint of what class you would use.

6
00:00:13,166 --> 00:00:17,233
Because I also would like you to practice
to research the documentation.

7
00:00:17,466 --> 00:00:20,733
So if I were to ask you to try to do this
before me, you know,

8
00:00:20,733 --> 00:00:24,500
to build and train this decision tree
regression model on this data set.

9
00:00:24,800 --> 00:00:28,200
Well, what you would have to do
would go to Google or Bing

10
00:00:28,200 --> 00:00:32,666
and then type in the search bar decision
tree regression class of scikit learn.

11
00:00:32,966 --> 00:00:34,533
Then you would find the name of the class.

12
00:00:34,533 --> 00:00:36,400
It would probably be on the first link.

13
00:00:36,400 --> 00:00:40,500
And then based on all the work we did
before you know, on how to build

14
00:00:40,500 --> 00:00:42,766
and train a machine
learning model with scikit learn,

15
00:00:42,766 --> 00:00:45,766
you would totally be able
to do this whole step on your own.

16
00:00:45,866 --> 00:00:46,200
All right.

17
00:00:46,200 --> 00:00:48,966
So if you want to do this, please
press pause on this video

18
00:00:48,966 --> 00:00:52,066
and do the exercise and I'm sure
you'll be proud to find the solution.

19
00:00:52,566 --> 00:00:53,100
All right.

20
00:00:53,100 --> 00:00:55,700
And now let's build a solution together.

21
00:00:55,700 --> 00:00:58,500
Starting with the creation of a new code
cell.

22
00:00:58,500 --> 00:00:59,600
And here we go.

23
00:00:59,600 --> 00:01:02,966
The name of the class
that can build a decision tree regression

24
00:01:02,966 --> 00:01:07,233
model in scikit
learn is called decision tree regressor.

25
00:01:07,633 --> 00:01:12,000
It is a class that belongs to the tree
module of scikit learn.

26
00:01:12,166 --> 00:01:16,133
And therefore here
we're going to start from scikit learn,

27
00:01:16,533 --> 00:01:20,333
from which
we're going to call that tree module

28
00:01:20,466 --> 00:01:24,233
and from which we're going to import
that decision.

29
00:01:24,833 --> 00:01:27,300
Google collab will guess it. There we go.

30
00:01:27,300 --> 00:01:30,066
So be careful.
It's not decision tree classifier.

31
00:01:30,066 --> 00:01:34,633
That will be for for three classification
but decision tree regressor.

32
00:01:34,900 --> 00:01:35,466
All right.

33
00:01:35,466 --> 00:01:37,800
This is based on the same model.
Decision trees.

34
00:01:37,800 --> 00:01:41,933
But decision tree regressor
will predict a continuous numerical value.

35
00:01:41,933 --> 00:01:44,933
And decision tree
classification will predict a category.

36
00:01:45,166 --> 00:01:50,100
All right then the next natural step here
is of course to create an object

37
00:01:50,266 --> 00:01:53,466
or an instance of that decision tree
regressor class.

38
00:01:53,700 --> 00:01:56,700
And we're going to call
that as usual regressor,

39
00:01:57,200 --> 00:02:01,700
which will be equal
of course to the call of this class.

40
00:02:01,700 --> 00:02:06,700
So I'm copying this and pasting that
right here and adding some parenthesis.

41
00:02:07,266 --> 00:02:07,566
All right.

42
00:02:07,566 --> 00:02:10,600
So now the question is do we have to input
anything here in the parenthesis.

43
00:02:10,800 --> 00:02:12,133
Well actually no you know

44
00:02:12,133 --> 00:02:16,133
there are not many parameters to tune
in the decision tree regression model.

45
00:02:16,233 --> 00:02:20,100
I don't recommend to spend too much time
tuning it and just to try it

46
00:02:20,200 --> 00:02:22,200
among your other regression models.

47
00:02:22,200 --> 00:02:26,166
But if you really want to tune it,
please note that there is this part

48
00:02:26,166 --> 00:02:30,233
ten of the course which covers
all the techniques of parameter tuning,

49
00:02:30,233 --> 00:02:34,533
which allows you to improve and optimize
the performance of a single model.

50
00:02:34,800 --> 00:02:37,566
And so don't worry,
you'll know how to deploy these techniques

51
00:02:37,566 --> 00:02:38,800
to and handsome model.

52
00:02:38,800 --> 00:02:39,366
But here

53
00:02:39,366 --> 00:02:43,133
we just want to learn how to build and
train the decision tree regression model.

54
00:02:43,133 --> 00:02:43,900
And therefore

55
00:02:43,900 --> 00:02:48,266
we will only input one parameter,
but which is just for training purposes.

56
00:02:48,300 --> 00:02:51,000
It is, you know,
that random state parameter

57
00:02:51,000 --> 00:02:54,433
which will allow us to get
the same result in the end, because indeed

58
00:02:54,666 --> 00:02:58,200
there are some random factors happening
when you build

59
00:02:58,200 --> 00:03:00,100
and train your decision tree regressor.

60
00:03:00,100 --> 00:03:03,966
And therefore if we don't fix the seed,
we will get slightly different

61
00:03:03,966 --> 00:03:05,033
results again.

62
00:03:05,033 --> 00:03:05,933
And you know, it's

63
00:03:05,933 --> 00:03:09,466
nicer to have the same results
so that we can all be on the same page.

64
00:03:09,466 --> 00:03:13,366
So we're just going to input random
underscore

65
00:03:13,366 --> 00:03:17,633
state parameter
and set that equal to zero.

66
00:03:17,700 --> 00:03:18,100
Right.

67
00:03:18,100 --> 00:03:22,600
We're fixing the seed here with that
zero value for the random state parameter.

68
00:03:23,133 --> 00:03:24,466
And now final step.

69
00:03:24,466 --> 00:03:26,966
Well you totally know how to finish this.

70
00:03:26,966 --> 00:03:32,100
We just need to take our regressor
and then call the fit method

71
00:03:32,400 --> 00:03:37,066
which takes as input of course
the matrix of features x the whole matrix

72
00:03:37,333 --> 00:03:40,700
and then the dependent variable
vector y'all.

73
00:03:40,700 --> 00:03:40,866
Right.

74
00:03:40,866 --> 00:03:45,133
So this will actually train
your decision tree regressor

75
00:03:45,466 --> 00:03:49,200
to understand the correlations
between the position levels here

76
00:03:49,400 --> 00:03:53,800
and the salaries,
after which you will have a trained model

77
00:03:54,133 --> 00:03:58,333
which you will be able to deploy
in production to predict a new result

78
00:03:58,333 --> 00:04:02,633
and especially that salary of the position
level 6.5.

79
00:04:03,000 --> 00:04:03,500
All right.

80
00:04:03,500 --> 00:04:05,033
So that's what
we'll do in the next tutorial.

81
00:04:05,033 --> 00:04:08,933
But first let's not forget to run the cell
to indeed

82
00:04:09,166 --> 00:04:12,166
build and train
that decision tree regression model.

83
00:04:12,300 --> 00:04:14,600
And also I want to say congratulations.

84
00:04:14,600 --> 00:04:17,600
If you, you know, press pause
at the beginning of this tutorial

85
00:04:17,700 --> 00:04:19,833
to code this by yourself first.

86
00:04:19,833 --> 00:04:23,100
And also congratulations
to those who tried because that's really

87
00:04:23,100 --> 00:04:23,600
what matters.

88
00:04:23,600 --> 00:04:26,733
You know, to take action
to at least try and practice.

89
00:04:26,966 --> 00:04:27,633
Okay.

90
00:04:27,633 --> 00:04:28,866
So all good here.

91
00:04:28,866 --> 00:04:31,200
That's great.
We have our regression model.

92
00:04:31,200 --> 00:04:35,800
And now in that same spirit of,
you know, taking action to practice

93
00:04:35,800 --> 00:04:39,900
and try to implement things on your own,
well, I would like you to try

94
00:04:39,900 --> 00:04:43,733
to predict that salary of the position
level number 6.5.

95
00:04:44,066 --> 00:04:47,700
And there is absolutely no trap here
or no difficulty.

96
00:04:47,700 --> 00:04:50,700
So I have no doubt
you will actually smash this.

97
00:04:50,700 --> 00:04:51,033
All right.

98
00:04:51,033 --> 00:04:55,266
So please try to do this
and will implement the easy solution

99
00:04:55,266 --> 00:04:57,400
together in the next tutorial.

100
00:04:57,400 --> 00:04:58,966
Until then, enjoy machine learning.
