1
00:00:00,033 --> 00:00:00,633
So there you go.

2
00:00:00,633 --> 00:00:03,566
That's the. Whole implementation.
Let me scroll down a bit.

3
00:00:03,566 --> 00:00:05,233
And as you can see. You will.

4
00:00:05,233 --> 00:00:06,300
Get many tools.

5
00:00:06,300 --> 00:00:10,033
So let's have a look at them
one by one through the table of contents.

6
00:00:10,500 --> 00:00:12,700
So the first thing I'll teach you is how.

7
00:00:12,700 --> 00:00:14,400
To import. The libraries.

8
00:00:14,400 --> 00:00:16,900
These are the libraries
we will always use.

9
00:00:16,900 --> 00:00:19,333
In any machine
learning model implementation.

10
00:00:19,333 --> 00:00:21,700
So we. Will include them in the template.

11
00:00:21,700 --> 00:00:23,066
So that. They can be ready to.

12
00:00:23,066 --> 00:00:25,333
Use. For our implementations.

13
00:00:25,333 --> 00:00:27,566
Then I. Will teach you how to. Import.

14
00:00:27,566 --> 00:00:31,033
The data set that exact same data set
which I.

15
00:00:31,033 --> 00:00:33,366
Just introduced to.
You a. Few seconds ago.

16
00:00:33,366 --> 00:00:36,466
That data set
I will show you how to first upload.

17
00:00:36,466 --> 00:00:40,000
It on Google Colab,
and then import it in your Python file.

18
00:00:40,333 --> 00:00:40,666
All right.

19
00:00:40,666 --> 00:00:43,333
So that's the second step
importing the dataset.

20
00:00:43,333 --> 00:00:45,466
Then I will teach. You how to take care.

21
00:00:45,466 --> 00:00:46,500
Of missing data.

22
00:00:46,500 --> 00:00:48,466
Because indeed in most of the.

23
00:00:48,466 --> 00:00:50,533
Data sets you work with in.

24
00:00:50,533 --> 00:00:54,000
Your machine learning career,
you might encounter some missing data.

25
00:00:54,300 --> 00:00:56,600
And that's the case here for our data set.

26
00:00:56,600 --> 00:01:00,600
As you can see, there is a missing salary
here and a missing age.

27
00:01:00,766 --> 00:01:02,866
And I will teach you. Exactly what to.

28
00:01:02,866 --> 00:01:04,466
Do in order to handle this.

29
00:01:04,466 --> 00:01:06,633
There are some techniques. Which are.

30
00:01:06,633 --> 00:01:08,400
The most relevant to then, you know,

31
00:01:08,400 --> 00:01:11,000
optimize the training of your machine
learning models.

32
00:01:11,000 --> 00:01:13,700
And I will. Show you the best of.
These techniques.

33
00:01:13,700 --> 00:01:15,066
All right then. After.

34
00:01:15,066 --> 00:01:17,800
Taking care of the missing. Data,
I will teach you how.

35
00:01:17,800 --> 00:01:18,900
To encode.

36
00:01:18,900 --> 00:01:21,300
Categorical data, whether it is the.

37
00:01:21,300 --> 00:01:22,333
Independent variable.

38
00:01:22,333 --> 00:01:24,700
Which is your predictors, or the.

39
00:01:24,700 --> 00:01:27,433
Dependent. Variable,
which is what you want to predict.

40
00:01:27,433 --> 00:01:29,833
So as you can see in our data.

41
00:01:29,833 --> 00:01:32,866
Set, well we have actually
two categorical variables.

42
00:01:32,866 --> 00:01:35,633
We have. This one
the country column containing in.

43
00:01:35,633 --> 00:01:37,066
These three categories.

44
00:01:37,066 --> 00:01:39,333
France, Spain and Germany as.

45
00:01:39,333 --> 00:01:40,833
Opposed. To all these numerical.

46
00:01:40,833 --> 00:01:42,666
Values. In the. Other variables.

47
00:01:42,666 --> 00:01:44,666
And also we have this categorical.

48
00:01:44,666 --> 00:01:46,500
Variable containing two. Categories.

49
00:01:46,500 --> 00:01:47,800
Yes and no.

50
00:01:47,800 --> 00:01:50,033
All right.
So I will teach you what to do with that.

51
00:01:50,033 --> 00:01:52,233
Situation so that then you can be ready.

52
00:01:52,233 --> 00:01:54,400
To preprocess. Any data.

53
00:01:54,400 --> 00:01:58,166
After this I will teach you how to split
the data set.

54
00:01:58,166 --> 00:02:00,033
Into the training set and a data set.

55
00:02:00,033 --> 00:02:03,566
And that step is very important
because each time you want to train a

56
00:02:03,566 --> 00:02:04,700
machine learning model.

57
00:02:04,700 --> 00:02:06,800
Well,
you have to create two separate sets.

58
00:02:06,800 --> 00:02:08,000
One training set.

59
00:02:08,000 --> 00:02:09,166
Where you're going to.

60
00:02:09,166 --> 00:02:12,633
Train your machine learning model to
understand the correlations inside your.

61
00:02:12,633 --> 00:02:13,366
Data set.

62
00:02:13,366 --> 00:02:14,266
And one test.

63
00:02:14,266 --> 00:02:17,333
Set that you're going to use
to evaluate your.

64
00:02:17,366 --> 00:02:18,266
Machine learning models.

65
00:02:18,266 --> 00:02:20,700
And that's therefore on new observations.

66
00:02:20,700 --> 00:02:21,666
Because the test set.

67
00:02:21,666 --> 00:02:25,300
You know, is like new data
on which the model wasn't trained.

68
00:02:25,300 --> 00:02:29,266
So that's very important to do this in
order to check that indeed there is not.

69
00:02:29,266 --> 00:02:30,033
Overfitting.

70
00:02:30,033 --> 00:02:32,633
You know, when your machine learning
model is trained too.

71
00:02:32,633 --> 00:02:34,266
Well. On the training set.

72
00:02:34,266 --> 00:02:35,666
So well that.

73
00:02:35,666 --> 00:02:38,100
It doesn't perform
well on new observations.

74
00:02:38,100 --> 00:02:40,166
And so that's. Why
this step. Is very important.

75
00:02:40,166 --> 00:02:43,166
And we will include it in the data
preprocessing template.

76
00:02:43,466 --> 00:02:44,133
And finally.

77
00:02:44,133 --> 00:02:45,000
I will teach you a.

78
00:02:45,000 --> 00:02:46,300
Very important tool

79
00:02:46,300 --> 00:02:50,233
that you might have to use in some of your
machine learning model implementations.

80
00:02:50,466 --> 00:02:52,466
Which is feature. Scaling.

81
00:02:52,466 --> 00:02:55,733
Which actually scales all your features.

82
00:02:55,733 --> 00:02:58,300
To make sure they are on the. Right scale.

83
00:02:58,300 --> 00:03:00,433
So you won't have to use. That
all the. Time.

84
00:03:00,433 --> 00:03:01,066
We will see.

85
00:03:01,066 --> 00:03:04,233
Of course, in each of the machine
learning model implementation, whether.

86
00:03:04,233 --> 00:03:06,300
We have to apply feature. Scaling or not.

87
00:03:06,300 --> 00:03:08,266
So you'll have. Everything, but.

88
00:03:08,266 --> 00:03:10,000
We. Have to include. This feature scaling.

89
00:03:10,000 --> 00:03:12,733
Tool in the data preprocessing toolkit,
because indeed.

90
00:03:12,733 --> 00:03:15,300
We'll have to use it. From time to time.

91
00:03:15,300 --> 00:03:15,900
All right.

92
00:03:15,900 --> 00:03:17,366
That's the table of contents.

93
00:03:17,366 --> 00:03:19,500
And we are going to reimplement.

94
00:03:19,500 --> 00:03:21,000
All. This from scratch.

95
00:03:21,000 --> 00:03:23,633
However, if you feel. You're comfortable.

96
00:03:23,633 --> 00:03:24,566
With all these tools.

97
00:03:24,566 --> 00:03:27,466
And you understand them already,
and you can't. Wait.

98
00:03:27,466 --> 00:03:30,466
To move on to the machine
learning implementations.

99
00:03:30,500 --> 00:03:33,033
Well, feel free to just read this code.

100
00:03:33,033 --> 00:03:34,433
And make sure you understand it.

101
00:03:34,433 --> 00:03:36,433
100%, and then move on to.

102
00:03:36,433 --> 00:03:37,000
Part two.

103
00:03:37,000 --> 00:03:38,333
Regression, where we're going to build.

104
00:03:38,333 --> 00:03:40,133
Our first regression. Models.

105
00:03:40,133 --> 00:03:43,933
However, I really insist
that this course must be action based.

106
00:03:43,933 --> 00:03:46,700
So I. Really want you to take action
as much as you can.

107
00:03:46,700 --> 00:03:47,533
And that's why.

108
00:03:47,533 --> 00:03:51,366
If you're ready to do this for data
preprocessing, well, stay.

109
00:03:51,366 --> 00:03:52,566
With me in this part one,

110
00:03:52,566 --> 00:03:56,333
because we are going to re-implement
each and every single one.

111
00:03:56,333 --> 00:03:57,400
Of these tools.

112
00:03:57,400 --> 00:03:58,366
And to do. This.

113
00:03:58,366 --> 00:03:59,400
Because remember that.

114
00:03:59,400 --> 00:04:00,733
This colab file is.

115
00:04:00,733 --> 00:04:02,633
Actually in read only mode.

116
00:04:02,633 --> 00:04:05,233
Well, to do
this, we'll have to create a copy by.

117
00:04:05,233 --> 00:04:06,466
Clicking. File here.

118
00:04:06,466 --> 00:04:09,266
And then. Save a copy in drive.

119
00:04:09,266 --> 00:04:09,866
This will.

120
00:04:09,866 --> 00:04:10,633
As you can see.

121
00:04:10,633 --> 00:04:12,966
Create a copy of this Google.

122
00:04:12,966 --> 00:04:14,700
Colab implementation on.

123
00:04:14,700 --> 00:04:16,500
Which you will be able to.

124
00:04:16,500 --> 00:04:17,833
Modify and.

125
00:04:17,833 --> 00:04:20,466
Mostly code. Your own implementation.

126
00:04:20,466 --> 00:04:20,866
All right.

127
00:04:20,866 --> 00:04:22,966
And that's exactly what we're going to do
now. We're going to.

128
00:04:22,966 --> 00:04:25,233
Actually. Re-Implement all. These code.

129
00:04:25,233 --> 00:04:26,666
Cells from scratch so that I make.

130
00:04:26,666 --> 00:04:28,933
Sure that you take. Action.

131
00:04:28,933 --> 00:04:31,133
And therefore we're. Going to remove.

132
00:04:31,133 --> 00:04:31,666
Each.

133
00:04:31,666 --> 00:04:32,266
One of.

134
00:04:32,266 --> 00:04:33,866
These code cells by.

135
00:04:33,866 --> 00:04:35,433
Clicking this trash button here.

136
00:04:35,433 --> 00:04:37,433
But make sure not to remove.

137
00:04:37,433 --> 00:04:38,533
Well, the text cells.

138
00:04:38,533 --> 00:04:39,300
This is a text cell.

139
00:04:39,300 --> 00:04:40,466
This is a code cell.

140
00:04:40,466 --> 00:04:42,433
So make sure to only remove.

141
00:04:42,433 --> 00:04:44,133
The code cells

142
00:04:44,133 --> 00:04:48,766
so that we can actually keep this
well highlighted structure of this.

143
00:04:48,766 --> 00:04:51,766
Implementation. All right. Almost done.

144
00:04:52,100 --> 00:04:54,000
A few code cells left.

145
00:04:54,000 --> 00:04:55,333
Trash button.

146
00:04:55,333 --> 00:04:57,333
Trash trash trash.

147
00:04:57,333 --> 00:04:59,866
And almost done. There we go.

148
00:04:59,866 --> 00:05:00,200
All right.

149
00:05:00,200 --> 00:05:03,700
So that's the whole structure
of this implementation.

150
00:05:03,900 --> 00:05:05,866
These are all the tools that you will.

151
00:05:05,866 --> 00:05:07,333
Probably need when.

152
00:05:07,333 --> 00:05:10,133
Preprocessing your future data
sets for your future.

153
00:05:10,133 --> 00:05:11,200
Machine learning models.

154
00:05:11,200 --> 00:05:14,766
So it's very important that you understand
well this implementation.

155
00:05:15,300 --> 00:05:17,566
So whenever you're. Ready let's start.

156
00:05:17,566 --> 00:05:19,133
From the next tutorial to.

157
00:05:19,133 --> 00:05:20,866
Master data preprocessing.