﻿1
00:00:00,100 --> 00:00:02,100
Hello and welcome to this tutorial.

2
00:00:02,100 --> 00:00:05,233
Now we are finally going to start
preparing the data

3
00:00:05,233 --> 00:00:08,233
so that our machine learning models
run correctly.

4
00:00:08,400 --> 00:00:11,200
And the first problem
that we have to deal with

5
00:00:11,200 --> 00:00:14,800
is the case where you have
some missing data in your data set.

6
00:00:15,433 --> 00:00:17,666
And that happens
quite a lot actually in real life.

7
00:00:17,666 --> 00:00:21,366
So you have to get the trick
to handle this problem

8
00:00:21,366 --> 00:00:24,833
and make it all good for your machine
learning model to run correctly.

9
00:00:25,266 --> 00:00:27,300
So let's have a look at the data set.

10
00:00:27,300 --> 00:00:30,300
Here I'm going to my Google Sheets tab.

11
00:00:30,800 --> 00:00:33,066
Okay so here is the data set.

12
00:00:33,066 --> 00:00:36,066
And as you can see
there is two missing data.

13
00:00:36,066 --> 00:00:39,233
There is one missing data
in the H column here for Spain

14
00:00:39,633 --> 00:00:43,166
and one missing data
in the salary column for Germany.

15
00:00:43,566 --> 00:00:45,833
So how can we handle this problem?

16
00:00:45,833 --> 00:00:48,833
Well the first idea is to remove

17
00:00:49,066 --> 00:00:52,433
the lines, the observations,
where there is some missing data.

18
00:00:52,466 --> 00:00:57,266
So what we could do is to remove this line
and remove this line.

19
00:00:57,633 --> 00:01:00,566
But that can be quite dangerous
because imagine

20
00:01:00,566 --> 00:01:03,566
this data set contains
crucial information.

21
00:01:03,866 --> 00:01:06,866
It would be quite dangerous
to remove an observation.

22
00:01:06,966 --> 00:01:10,933
So we need to figure out a better idea
to handle this problem.

23
00:01:11,533 --> 00:01:12,366
And another idea.

24
00:01:12,366 --> 00:01:15,500
And that's actually the most common idea
to handle missing

25
00:01:15,500 --> 00:01:19,100
data is to take the mean of the columns.

26
00:01:19,466 --> 00:01:22,866
So here we are going to replace this
missing data here

27
00:01:23,100 --> 00:01:26,100
by the mean of all the values
in the column h.

28
00:01:26,466 --> 00:01:29,500
And that's the same for every feature
that contains missing data.

29
00:01:29,866 --> 00:01:31,433
We replace this missing data

30
00:01:31,433 --> 00:01:35,100
by the mean of the values in the column
that contains this missing data.

31
00:01:35,666 --> 00:01:37,500
Okay. So let's do this.

32
00:01:37,500 --> 00:01:40,433
First we need to take our data set.

33
00:01:40,433 --> 00:01:41,766
So data set.

34
00:01:41,766 --> 00:01:45,100
Let's start by taking care
of the missing value in the h column.

35
00:01:45,500 --> 00:01:47,766
So here we will need to take the h column.

36
00:01:47,766 --> 00:01:51,133
And to do this in R
we need to add a dollar sign here.

37
00:01:51,566 --> 00:01:53,466
And here we choose h.

38
00:01:53,466 --> 00:01:59,000
So by doing that data set dollar h we're
taking the column age of the data set.

39
00:01:59,700 --> 00:02:01,366
Then we're going to add equals.

40
00:02:01,366 --> 00:02:03,866
And then we're going to add an if else.

41
00:02:03,866 --> 00:02:07,866
So I'm going to type
if else here then parentheses.

42
00:02:09,000 --> 00:02:11,566
In the ifelse function
you have to input three parameters.

43
00:02:11,566 --> 00:02:13,600
The first parameter is your condition.

44
00:02:13,600 --> 00:02:18,100
If condition the second parameter is
the value you want to input.

45
00:02:18,100 --> 00:02:19,566
If the condition is true

46
00:02:19,566 --> 00:02:22,600
and this is the value you want to input
if the condition is false.

47
00:02:23,100 --> 00:02:25,033
So let's first input the condition.

48
00:02:25,033 --> 00:02:29,100
The condition is going to be is.na
parentheses.

49
00:02:29,100 --> 00:02:32,100
Data set dollar age.

50
00:02:32,933 --> 00:02:34,800
And that's it

51
00:02:34,800 --> 00:02:37,033
is Na is a function that tells

52
00:02:37,033 --> 00:02:40,033
if the value in the function
is missing or not.

53
00:02:40,133 --> 00:02:43,200
So by putting is Na data set dollars age.

54
00:02:43,333 --> 00:02:47,766
We are checking to see if all the values
in the column age are missing.

55
00:02:48,066 --> 00:02:52,800
So this will return true
if the value in the column age is missing

56
00:02:53,066 --> 00:02:56,200
and false if the value in the column
h is not missing.

57
00:02:56,566 --> 00:02:58,300
Okay, so that's the condition.

58
00:02:58,300 --> 00:03:02,366
And now the second parameter is the value
that is going to be returned.

59
00:03:02,400 --> 00:03:04,866
If the condition above is true.

60
00:03:04,866 --> 00:03:08,233
And of course if the condition is true
that means that there is a missing value.

61
00:03:08,400 --> 00:03:10,966
And that means that we have to replace it
with the average.

62
00:03:10,966 --> 00:03:12,300
So here we're going to impute

63
00:03:12,300 --> 00:03:15,633
as the second parameter
the average of the column age.

64
00:03:16,133 --> 00:03:16,466
Okay.

65
00:03:16,466 --> 00:03:19,033
And to compute the average in R
there is a simple way.

66
00:03:19,033 --> 00:03:22,366
We can type RV then data set

67
00:03:23,433 --> 00:03:24,133
dollar sign

68
00:03:24,133 --> 00:03:27,133
age because
we want to take the column age.

69
00:03:27,500 --> 00:03:29,800
Then comma.

70
00:03:29,800 --> 00:03:31,566
And then here
we're going to add a function.

71
00:03:31,566 --> 00:03:35,066
So we're going to type fund in capitals
phone equals

72
00:03:35,400 --> 00:03:38,300
then function then function x.

73
00:03:38,300 --> 00:03:39,966
This is still part of the R syntax.

74
00:03:39,966 --> 00:03:43,166
We're just making a function here
which is going to be the mean function.

75
00:03:43,533 --> 00:03:46,533
And then we have to specify
what this function will be.

76
00:03:46,600 --> 00:03:51,000
And so this function is of course the mean
which is an existing function in R.

77
00:03:51,600 --> 00:03:55,000
So here parenthesis x comma.

78
00:03:55,000 --> 00:03:58,400
And here we're just going to add na dot

79
00:03:58,633 --> 00:04:01,633
rmm equals true.

80
00:04:02,466 --> 00:04:04,000
And that means that we ask.

81
00:04:04,000 --> 00:04:07,433
And that means that we ask R
to include the missing values.

82
00:04:07,533 --> 00:04:11,333
When R will go through the whole column
aid to compute the mean of the values.

83
00:04:11,600 --> 00:04:15,433
And then we still need to close
the parenthesis here again then comma.

84
00:04:15,466 --> 00:04:17,633
So that's it for the second parameter.

85
00:04:17,633 --> 00:04:20,366
And now we need to add
the third parameter.

86
00:04:20,366 --> 00:04:22,300
And in your opinion
what is it going to be.

87
00:04:23,366 --> 00:04:26,300
Well this third parameter is the value
you want to return.

88
00:04:26,300 --> 00:04:28,833
If the first condition is not true.

89
00:04:28,833 --> 00:04:31,933
That means that the value in the column
age is not missing.

90
00:04:31,966 --> 00:04:34,333
So that means that the value exists.

91
00:04:34,333 --> 00:04:37,633
So here we are
simply going to put data set

92
00:04:39,533 --> 00:04:41,466
age okay.

93
00:04:41,466 --> 00:04:45,333
And that's done by typing this
we replace the missing value

94
00:04:45,333 --> 00:04:48,466
in the column age
by the mean of the column age itself.

95
00:04:48,900 --> 00:04:51,900
So let's select
this and let's see what happens.

96
00:04:52,100 --> 00:04:54,533
So command and control
plus enter to execute.

97
00:04:54,533 --> 00:04:57,533
And here it is and executed properly.

98
00:04:57,600 --> 00:05:00,600
And now let's look at our data set
by clicking on this tab here.

99
00:05:01,266 --> 00:05:02,233
And good perfect.

100
00:05:02,233 --> 00:05:05,200
The missing value that was here in the age
column was replaced

101
00:05:05,200 --> 00:05:07,933
by the mean of the column. Great.

102
00:05:07,933 --> 00:05:11,066
So now let's do the same for the salary.

103
00:05:11,633 --> 00:05:14,766
We're just going to copy this paste.

104
00:05:15,300 --> 00:05:19,233
And we're just going to replace age
by the salary here.

105
00:05:20,000 --> 00:05:20,800
Here as well.

106
00:05:21,900 --> 00:05:23,700
Here as well.

107
00:05:23,700 --> 00:05:26,700
And also here.

108
00:05:26,800 --> 00:05:27,633
Great.

109
00:05:27,633 --> 00:05:29,933
And now we have to be careful
with something.

110
00:05:29,933 --> 00:05:31,366
It has to be a line here.

111
00:05:31,366 --> 00:05:33,633
So we just have to do this.

112
00:05:33,633 --> 00:05:35,633
And same for here.

113
00:05:35,633 --> 00:05:37,633
This okay. And now we're fine.

114
00:05:37,633 --> 00:05:40,566
Now we're ready to select this code
section here.

115
00:05:40,566 --> 00:05:42,933
Press Command and Control
plus enter to execute.

116
00:05:42,933 --> 00:05:43,566
Here it is.

117
00:05:43,566 --> 00:05:46,233
Let's check our data set and perfect.

118
00:05:46,233 --> 00:05:49,266
The missing value in the salary
that was here was replaced

119
00:05:49,266 --> 00:05:54,900
by the mean of the salary column $63,777.
