﻿1
00:00:00,133 --> 00:00:01,100
Hello my friends.

2
00:00:01,100 --> 00:00:05,233
Welcome to this new section on now
Decision Tree Regression.

3
00:00:05,233 --> 00:00:09,333
So we're about to tackle
a new practical activity where we will

4
00:00:09,333 --> 00:00:12,766
all learn together how to build
the decision tree regression model.

5
00:00:13,166 --> 00:00:17,100
You will see that it will be super easy
based on what we did before.

6
00:00:17,100 --> 00:00:22,200
You know, with all the feature scaling
and inverse transformation of the SVR, now

7
00:00:22,200 --> 00:00:26,633
we will not have to apply feature scaling
and therefore we will just smash this.

8
00:00:26,633 --> 00:00:27,366
All right.

9
00:00:27,366 --> 00:00:28,033
Are you ready?

10
00:00:28,033 --> 00:00:29,133
Are you ready to start.

11
00:00:29,133 --> 00:00:32,133
Before we go into this part two regression
folder,

12
00:00:32,200 --> 00:00:35,200
let's just make sure that everyone here
is on the same page.

13
00:00:35,333 --> 00:00:38,033
I give you the link to this folder
just before this tutorial.

14
00:00:38,033 --> 00:00:39,966
So you just have to click the link.

15
00:00:39,966 --> 00:00:42,133
And then now
we should all be on the same page.

16
00:00:42,133 --> 00:00:45,133
So we're going to go into part
two regression.

17
00:00:45,300 --> 00:00:48,300
And then section eight decision tree
regression.

18
00:00:48,400 --> 00:00:50,766
We are almost at the end of the regression
part.

19
00:00:50,766 --> 00:00:54,266
Congratulations for the great progress
you've made so far.

20
00:00:54,533 --> 00:00:55,300
And now we're going to go,

21
00:00:55,300 --> 00:00:59,833
of course, to Python in order
to fine our files for this section.

22
00:01:00,300 --> 00:01:01,800
So there are two files here.

23
00:01:01,800 --> 00:01:06,566
The Python implementation of the decision
tree regression model in Ipynb format,

24
00:01:06,566 --> 00:01:10,033
which you can open with either
Google Colab or Jupyter Notebook.

25
00:01:10,333 --> 00:01:13,500
And our same position salaries data set

26
00:01:13,800 --> 00:01:17,333
containing that same data
of the previous company, showing the

27
00:01:17,333 --> 00:01:21,966
different position levels from 1 to 10,
corresponding to business analyst to CEO,

28
00:01:22,066 --> 00:01:27,633
and the corresponding salaries
from 45,000 to $1 million per year.

29
00:01:28,100 --> 00:01:28,500
All right.

30
00:01:28,500 --> 00:01:32,100
So this time we're going to train
a decision tree regression model

31
00:01:32,100 --> 00:01:35,700
to understand the correlations
between these two features.

32
00:01:35,900 --> 00:01:38,700
However,
I have to say something important here.

33
00:01:38,700 --> 00:01:43,066
The decision tree regression
model is not really well adapted to,

34
00:01:43,200 --> 00:01:45,300
you know, these simple data sets,

35
00:01:45,300 --> 00:01:48,566
you know, with only one feature
and the dependent variable vector.

36
00:01:48,966 --> 00:01:53,166
You'll see what I mean by that at the end,
you know, on the visualization graphs.

37
00:01:53,400 --> 00:01:57,200
But having said that,
I would like you not to worry

38
00:01:57,433 --> 00:02:00,666
because to implement
one of the decision tree regression model

39
00:02:00,666 --> 00:02:05,233
we were about to build
will still work on any other data sets.

40
00:02:05,233 --> 00:02:06,933
You know, with several features.

41
00:02:06,933 --> 00:02:10,066
Here we have one feature, but the code
we were about to make will work

42
00:02:10,066 --> 00:02:13,500
for data
set having any number of features.

43
00:02:13,766 --> 00:02:14,100
All right.

44
00:02:14,100 --> 00:02:17,400
So even if the results
won't be beautiful in the end, well,

45
00:02:17,400 --> 00:02:21,366
you will still be able to use this
decision tree regression implementation

46
00:02:21,600 --> 00:02:24,900
on your other data sets,
even if they have hundreds of features.

47
00:02:24,900 --> 00:02:29,266
But then make sure to add some data
preprocessing tools if needed.

48
00:02:29,400 --> 00:02:32,566
For example,
if your data set has some categorical data

49
00:02:32,566 --> 00:02:35,866
or missing data, but
you don't have to apply feature scaling

50
00:02:35,866 --> 00:02:39,900
for decision tree regression,
and neither for random forest regression,

51
00:02:39,900 --> 00:02:42,900
which will be our next model,
you know, in the next section.

52
00:02:43,500 --> 00:02:44,000
All right.

53
00:02:44,000 --> 00:02:46,666
So that's the important thing
I wanted to say here.

54
00:02:46,666 --> 00:02:51,500
And now we're going to start
our implementation by double clicking

55
00:02:51,500 --> 00:02:55,900
this file here which you can either open
with Google Collab

56
00:02:55,900 --> 00:02:59,566
if you like, Google Collab
like me or Jupyter Notebook.

57
00:02:59,700 --> 00:03:00,033
All right.

58
00:03:00,033 --> 00:03:01,600
So choose your favorite.

59
00:03:01,600 --> 00:03:04,600
And now let's open Google Colaboratory.

60
00:03:04,633 --> 00:03:08,166
It is opening the notebook and

61
00:03:09,266 --> 00:03:09,966
here we go.

62
00:03:09,966 --> 00:03:12,266
That's the whole implementation.

63
00:03:12,266 --> 00:03:12,566
All right.

64
00:03:12,566 --> 00:03:16,600
So as usual now we're going to create
a copy of this notebook.

65
00:03:16,600 --> 00:03:21,600
Because this is in read only mode
which means you can't modify it or recode

66
00:03:21,600 --> 00:03:25,900
it. So we're going to go to file here
and then click here.

67
00:03:26,100 --> 00:03:28,366
Save a copy in drive.

68
00:03:28,366 --> 00:03:31,433
And this will as you can see
create a copy of this notebook

69
00:03:31,766 --> 00:03:35,333
on which you will be able to recode on it.

70
00:03:35,333 --> 00:03:38,600
You know, re-implement
this decision tree regression model.

71
00:03:39,100 --> 00:03:39,933
Perfect.

72
00:03:39,933 --> 00:03:42,266
So now you know the next step.

73
00:03:42,266 --> 00:03:44,633
We're going to delete the code cells.

74
00:03:44,633 --> 00:03:50,700
But since it is the third time we actually
work on this position salaries data set.

75
00:03:50,966 --> 00:03:55,100
And of course each time
it's the same two first steps of the data

76
00:03:55,100 --> 00:03:58,266
preprocessing phase importing
the libraries and importing the data sets.

77
00:03:58,466 --> 00:04:00,900
Well,
this time we won't re-implement this.

78
00:04:00,900 --> 00:04:03,500
We will just leave them
and not delete them.

79
00:04:03,500 --> 00:04:06,766
So we will just delete
all the code cells from here.

80
00:04:06,766 --> 00:04:08,466
You know, from the step training,
the decision tree

81
00:04:08,466 --> 00:04:10,600
regression model on the whole data set.
All right.

82
00:04:10,600 --> 00:04:11,533
So let's do this.

83
00:04:11,533 --> 00:04:15,733
Let's start by deleting this one
because we will re-implement it together

84
00:04:15,966 --> 00:04:17,366
then this one.

85
00:04:17,366 --> 00:04:19,300
And now this one.

86
00:04:19,300 --> 00:04:20,133
All right.

87
00:04:20,133 --> 00:04:20,766
Perfect.

88
00:04:20,766 --> 00:04:24,433
Also you can notice that at the end
we will only visualize

89
00:04:24,433 --> 00:04:27,433
the decision tree
regression results in high resolution.

90
00:04:27,433 --> 00:04:30,033
Because you will see
and I will show this to you

91
00:04:30,033 --> 00:04:33,533
that the decision tree regression
results in low resolution.

92
00:04:33,533 --> 00:04:37,733
You know, without applying the grid
solution will absolutely not make sense.

93
00:04:37,733 --> 00:04:40,200
And I will explain that
at the end of this section.
