1
00:00:00,533 --> 00:00:01,333
Hello and

2
00:00:01,333 --> 00:00:05,300
welcome to the final tutorial of part
one Data Pre-processing.

3
00:00:05,600 --> 00:00:10,766
We finally completed all the steps
that there is to do to prepare any data

4
00:00:10,766 --> 00:00:13,766
sets on which we will build our machine
learning models.

5
00:00:13,800 --> 00:00:14,400
And now what

6
00:00:14,400 --> 00:00:17,933
we only need to do left is to prepare
the data pre-processing template.

7
00:00:18,200 --> 00:00:21,433
Because we learned a few stuff,
we learned how to import a data

8
00:00:21,433 --> 00:00:25,766
set to take care of missing data,
to encode categorical data,

9
00:00:26,000 --> 00:00:30,066
to split the data set into the training
set and the data set, and to apply feature

10
00:00:30,066 --> 00:00:33,066
scaling to put all of our variables
on the same scale.

11
00:00:33,433 --> 00:00:35,566
However, in the data
pre-processing template,

12
00:00:35,566 --> 00:00:37,533
we're not going to include all of these.

13
00:00:37,533 --> 00:00:40,600
We are only going to include the importing
the libraries.

14
00:00:41,066 --> 00:00:43,733
Then of course
we need to import the data set.

15
00:00:43,733 --> 00:00:46,866
Then regarding missing data,
I just wanted to show you how to take care

16
00:00:46,866 --> 00:00:49,866
of that in case you have some missing data
in your data sets.

17
00:00:50,166 --> 00:00:51,333
You know, in your work.

18
00:00:51,333 --> 00:00:53,700
So we are not going to include this
in the templates.

19
00:00:53,700 --> 00:00:55,800
It's just good to know
how to take care of this.

20
00:00:55,800 --> 00:00:58,533
But then we will focus on the machine
learning models themselves.

21
00:00:58,533 --> 00:01:02,366
And if in your work experience
you encounter missing data

22
00:01:02,366 --> 00:01:03,766
you know how to handle this.

23
00:01:03,766 --> 00:01:06,066
Any missing data issues then?

24
00:01:06,066 --> 00:01:09,600
As for categorical data,
we won't include it either in the template

25
00:01:09,600 --> 00:01:11,100
because we are going to find

26
00:01:11,100 --> 00:01:15,300
very few examples of data
where we have to encode the data.

27
00:01:15,300 --> 00:01:16,833
There will be some examples,

28
00:01:16,833 --> 00:01:20,333
but we won't include it in the templates
because our data sets are going

29
00:01:20,333 --> 00:01:21,300
to be well prepared

30
00:01:21,300 --> 00:01:25,566
so that we can mainly focus on the machine
learning models and get the maximum fun.

31
00:01:26,200 --> 00:01:29,300
And then of course,
we will include this in the template,

32
00:01:29,300 --> 00:01:32,300
splitting the data sets
into the training set and the test set.

33
00:01:32,400 --> 00:01:34,300
Because that's a very important step.

34
00:01:34,300 --> 00:01:35,600
You need to split your data

35
00:01:35,600 --> 00:01:40,233
set between training and test,
because you need to evaluate your model

36
00:01:40,233 --> 00:01:45,700
on a different set than the sets on which
you build your model and feature scaling.

37
00:01:45,700 --> 00:01:47,066
Okay, so feature scaling.

38
00:01:47,066 --> 00:01:50,966
I hesitated to include it in the template,
but we are going to include it

39
00:01:51,233 --> 00:01:54,933
only we're going to put that as comment
because you're going to see that

40
00:01:55,266 --> 00:01:59,366
there are several libraries in R
and Python, and some of them require us

41
00:01:59,366 --> 00:02:03,000
to apply features getting to the data,
and some of them don't.

42
00:02:03,400 --> 00:02:05,100
So most of them don't.

43
00:02:05,100 --> 00:02:07,666
Actually most of them
take care of that for you.

44
00:02:07,666 --> 00:02:09,300
You don't have to do it manually,

45
00:02:09,300 --> 00:02:12,600
but you will see that some libraries
don't apply the feature scaling.

46
00:02:12,933 --> 00:02:13,500
I won't tell you

47
00:02:13,500 --> 00:02:17,066
which models there are right now, because
I will let you find out the surprise.

48
00:02:17,266 --> 00:02:18,900
But keep that in mind

49
00:02:18,900 --> 00:02:22,100
and you will see that sometimes
we will have to use feature scaling.

50
00:02:22,666 --> 00:02:24,600
Okay, so let's start making the template.

51
00:02:24,600 --> 00:02:25,966
It's going to be very quick.

52
00:02:25,966 --> 00:02:28,300
Let's do it right now. Let's jump to R.

53
00:02:28,300 --> 00:02:28,733
All right.

54
00:02:28,733 --> 00:02:31,600
So here is all the steps
that we did together.

55
00:02:31,600 --> 00:02:33,166
So importing the data sets.

56
00:02:33,166 --> 00:02:36,933
Yes we're keeping it
of course taking care of missing data.

57
00:02:36,933 --> 00:02:40,433
That was just to show you
we won't need it in the future. So.

58
00:02:40,633 --> 00:02:42,766
But you will probably need it
in your future.

59
00:02:42,766 --> 00:02:45,300
But in the future of this course,
we won't need it.

60
00:02:45,300 --> 00:02:46,300
So we will remove it

61
00:02:47,400 --> 00:02:48,733
and coding categorical data.

62
00:02:48,733 --> 00:02:49,466
Same here.

63
00:02:49,466 --> 00:02:53,400
We will have to do it once or twice,
but not all the time.

64
00:02:53,400 --> 00:02:54,666
So we will remove it.

65
00:02:54,666 --> 00:02:57,500
And as I told you, I'm
putting this in a separate file

66
00:02:57,500 --> 00:03:00,500
so that you can still use it
if you need it for your work.

67
00:03:01,866 --> 00:03:03,166
All right.

68
00:03:03,166 --> 00:03:05,100
And now I think we have everything

69
00:03:05,100 --> 00:03:07,533
splitting the data
set into the training set and test set.

70
00:03:07,533 --> 00:03:12,300
Of course we are keeping that and feature
scaling feature scaling.

71
00:03:12,700 --> 00:03:15,166
As for Python
we're going to put that in comment.

72
00:03:15,166 --> 00:03:18,066
So then the three double quotes

73
00:03:18,066 --> 00:03:21,066
is not a way to put multiline
comments in R.

74
00:03:21,366 --> 00:03:22,866
So unfortunately we cannot do it.

75
00:03:22,866 --> 00:03:25,466
What we need to do is
this is a good trick.

76
00:03:25,466 --> 00:03:29,900
You select the two lines here
and then you press Command and Control

77
00:03:30,266 --> 00:03:33,266
plus shift plus C.

78
00:03:33,766 --> 00:03:36,200
And that puts all your lines in comment.

79
00:03:36,200 --> 00:03:39,500
And eventually we'll just add
a final touch to this template,

80
00:03:39,700 --> 00:03:44,133
which is the case where we need
to take a subset of our data set.

81
00:03:44,533 --> 00:03:47,633
And in that case
we will add the line data set

82
00:03:48,600 --> 00:03:49,666
equals.

83
00:03:49,666 --> 00:03:52,666
Then we take the same data set.

84
00:03:53,366 --> 00:03:56,133
And we use the square brackets to select

85
00:03:56,133 --> 00:03:59,200
the indexes of the columns
we are interested in to build our model.

86
00:03:59,466 --> 00:04:01,833
So let's just put some numbers here.

87
00:04:01,833 --> 00:04:04,833
And if we need to select
specific columns of interest

88
00:04:04,833 --> 00:04:08,100
in our future data sets
we will change those indexes.

89
00:04:08,633 --> 00:04:11,633
But for now let's
just put this line in comment.

90
00:04:13,233 --> 00:04:14,500
All right.

91
00:04:14,500 --> 00:04:15,000
All right.

92
00:04:15,000 --> 00:04:17,900
So the template is ready on as well.

93
00:04:17,900 --> 00:04:20,166
We are ready to start our machine
learning models.

94
00:04:20,166 --> 00:04:22,266
You have no idea how I'm excited
about that.

95
00:04:22,266 --> 00:04:24,600
I can't wait to show you
how it's going to be a lot of fun.

96
00:04:24,600 --> 00:04:26,766
We're going to make awesome predictions.

97
00:04:26,766 --> 00:04:27,366
You know, we will.

98
00:04:27,366 --> 00:04:27,900
We will have

99
00:04:27,900 --> 00:04:31,500
some kind of scenarios, business problems,
and we will have to solve them.

100
00:04:31,666 --> 00:04:34,700
And then we will use the power of machine
learning to solve these problems.

101
00:04:35,033 --> 00:04:37,966
You will see how machine learning is
powerful.

102
00:04:37,966 --> 00:04:40,966
It gives you amazing
predictions, amazing results.

103
00:04:41,000 --> 00:04:44,466
And keep in mind you will never be lost
because we will always know what

104
00:04:44,466 --> 00:04:46,766
we are coding.
We will always visualize it.

105
00:04:46,766 --> 00:04:49,500
There will always be some visual graphics

106
00:04:49,500 --> 00:04:52,533
of what we are doing
when we make every machine learning model.

107
00:04:52,833 --> 00:04:55,166
So I can't wait to start this with you.

108
00:04:55,166 --> 00:04:58,000
This was the boring part,
but the very important part to do

109
00:04:58,000 --> 00:05:01,400
so that's great that we did it
and now it's time to have fun.

110
00:05:02,033 --> 00:05:05,400
So thank you for watching this data
pre-processing tutorials.

111
00:05:05,400 --> 00:05:08,100
Congratulations for completing this part.

112
00:05:08,100 --> 00:05:10,433
You are now ready to make machine
learning models,

113
00:05:10,433 --> 00:05:13,033
and I look forward
to seeing you in the next part.

114
00:05:13,033 --> 00:05:14,833
Until then, enjoy machine learning.