﻿1
00:00:00,033 --> 00:00:01,500
Okay, then.

2
00:00:01,500 --> 00:00:02,733
Almost last step.

3
00:00:02,733 --> 00:00:05,233
Now remember, this is just an object.

4
00:00:05,233 --> 00:00:08,700
We haven't connected anything yet
to our matrix of features.

5
00:00:08,966 --> 00:00:14,100
So the next step is indeed to apply this
input object on the matrix of features.

6
00:00:14,600 --> 00:00:15,933
So how are we going to do that.

7
00:00:15,933 --> 00:00:20,066
Well remember that a class contains
an assemble of instructions

8
00:00:20,200 --> 00:00:24,166
but also some operations and actions
which you can apply

9
00:00:24,400 --> 00:00:26,433
to other objects or variables.

10
00:00:26,433 --> 00:00:27,933
And these are called methods.

11
00:00:27,933 --> 00:00:29,566
You know they're like functions.

12
00:00:29,566 --> 00:00:32,566
And one of them is exactly the fit method.

13
00:00:32,800 --> 00:00:38,300
The fit method will exactly connect
this input to the matrix of features.

14
00:00:38,333 --> 00:00:41,433
In other words,
what this fit method will do is it will

15
00:00:41,433 --> 00:00:45,033
look at the missing values
in, you know, the salary column.

16
00:00:45,300 --> 00:00:48,400
And also it will compute
the average of the salaries.

17
00:00:48,700 --> 00:00:51,000
All right.
So that that simply what it will do.

18
00:00:51,000 --> 00:00:55,100
And then that will not be enough to do the
and we want to do

19
00:00:55,100 --> 00:00:59,233
the replacement will have to call
another method called transform time

20
00:00:59,400 --> 00:01:02,700
and which will this time apply
the transformation meaning

21
00:01:02,966 --> 00:01:06,866
it will replace the missing salary here
by the average of the salaries.

22
00:01:07,100 --> 00:01:09,400
Okay, so let's do this.

23
00:01:09,400 --> 00:01:11,966
Let's first call the fit method.

24
00:01:11,966 --> 00:01:16,066
In order to do this well of course
we have to call first our object Imputer.

25
00:01:16,733 --> 00:01:19,366
And then from this object
you know adding a dot,

26
00:01:19,366 --> 00:01:22,366
we will call the fit method

27
00:01:22,600 --> 00:01:26,266
which has some parentheses because
it's like a function inside a class.

28
00:01:26,300 --> 00:01:29,066
And what does this function expect
as arguments.

29
00:01:29,066 --> 00:01:32,033
Well,
it simply expects all the columns of X

30
00:01:32,033 --> 00:01:35,900
with numerical values,
but only the ones with numerical values.

31
00:01:35,900 --> 00:01:38,900
Not the ones with text or strings
categories.

32
00:01:39,000 --> 00:01:41,033
And so. Well, how do we get this column?

33
00:01:41,033 --> 00:01:42,133
Well first let's get

34
00:01:42,133 --> 00:01:46,066
the matrix of features X because that's
where we want to replace the missing data.

35
00:01:46,066 --> 00:01:48,600
And from this matrix of features x.

36
00:01:48,600 --> 00:01:51,300
Well first we're going to
look at all the rows.

37
00:01:51,300 --> 00:01:52,433
You know this fit

38
00:01:52,433 --> 00:01:57,166
method will read the whole column
that we specify inside this method.

39
00:01:57,466 --> 00:02:01,066
But then for the columns here
you know we could specify all the columns

40
00:02:01,300 --> 00:02:03,833
where to look for some missing data.

41
00:02:03,833 --> 00:02:06,300
However, this first column has danger.

42
00:02:06,300 --> 00:02:09,700
You know it is a column with strings,
and therefore this might cause

43
00:02:09,700 --> 00:02:12,700
a warning or an error
when looking for some missing data here.

44
00:02:12,833 --> 00:02:17,000
Therefore, we're only going to specify
these columns with only real numbers

45
00:02:17,133 --> 00:02:18,666
age and salary.

46
00:02:18,666 --> 00:02:23,533
And therefore here we're going
to enter the range from one to be careful

47
00:02:23,533 --> 00:02:27,600
not to because remember the upper
bound of a range in Python is excluded.

48
00:02:27,600 --> 00:02:30,400
So if we exclude two
this will list through the salary.

49
00:02:30,400 --> 00:02:33,533
Therefore we have to go up to three okay.

50
00:02:33,800 --> 00:02:34,366
Up to three.

51
00:02:34,366 --> 00:02:36,733
So that well this fit method

52
00:02:36,733 --> 00:02:40,933
will look for all the missing values
in the h column and the salary column.

53
00:02:40,933 --> 00:02:44,700
So here we're specifying specific columns
which are the h column in

54
00:02:44,700 --> 00:02:45,700
the salary column.

55
00:02:45,700 --> 00:02:48,600
And that's because we know
that there is a missing salary.

56
00:02:48,600 --> 00:02:51,000
And by the way there is also a missing H.

57
00:02:51,000 --> 00:02:55,933
However, I recommend on a general rule
to select all the numerical columns,

58
00:02:56,100 --> 00:02:59,100
because in your career
you will actually work with huge data sets

59
00:02:59,300 --> 00:03:02,233
and you won't be able to see where
the missing values are.

60
00:03:02,233 --> 00:03:07,600
So just include all the numerical columns
to make sure to replace any missing data.

61
00:03:07,966 --> 00:03:11,100
Remember to exclude these ones
the string columns okay,

62
00:03:11,400 --> 00:03:14,400
so that's
just something important to consider.

63
00:03:14,433 --> 00:03:15,466
And then here we go.

64
00:03:15,466 --> 00:03:19,200
This will connect our imputer
to our matrix of features.

65
00:03:19,200 --> 00:03:23,666
And now final step
we have to call the transform method

66
00:03:23,866 --> 00:03:27,133
once again from our Imputer object.

67
00:03:27,933 --> 00:03:32,233
And so this transform
method will exactly do that

68
00:03:32,233 --> 00:03:36,900
replacement of the missing salary here
by the mean of the salaries.

69
00:03:37,300 --> 00:03:38,766
And same for the missing age.

70
00:03:38,766 --> 00:03:42,833
It will be replaced by the mean
of all the ages in the H column.

71
00:03:43,400 --> 00:03:46,400
And so according to you,
what do we have to input here?

72
00:03:46,533 --> 00:03:48,366
Well, there is no trap here.

73
00:03:48,366 --> 00:03:53,533
We of course have to input the columns of
X where we want to replace missing data.

74
00:03:53,533 --> 00:03:57,000
And so these are the h column
and the salary column.

75
00:03:57,233 --> 00:04:00,066
And therefore we simply have
to input exactly

76
00:04:00,066 --> 00:04:03,066
the same as
what was input in the fit method.

77
00:04:03,066 --> 00:04:08,366
So we just going to take this copy this and paste that inside the transform method.

78
00:04:09,400 --> 00:04:09,933
However be

79
00:04:09,933 --> 00:04:15,166
careful this transform method
actually returns the new updated version

80
00:04:15,166 --> 00:04:16,400
of the matrix of features

81
00:04:16,400 --> 00:04:20,700
x with the two replacements
of the missing salary and the missing H,

82
00:04:21,033 --> 00:04:24,033
and therefore what we want to do now,
and that's the last thing we have to do,

83
00:04:24,133 --> 00:04:29,200
is to indeed update our matrix of features
X, and to do this well, since

84
00:04:29,500 --> 00:04:33,866
this exactly returns these two columns
here, with that replacement done

85
00:04:34,166 --> 00:04:38,766
well, what we want to do to update
X is actually to take this,

86
00:04:38,766 --> 00:04:43,200
you know, take the second and third column
of X matrix features and

87
00:04:44,266 --> 00:04:48,333
change it by what will be returned
by this transform function

88
00:04:48,600 --> 00:04:52,600
of the input object,
so that the second and third columns of x

89
00:04:52,733 --> 00:04:56,266
will be replaced
with that average age and average salary.

90
00:04:56,400 --> 00:04:59,833
And therefore the whole matrix of features
X will be exactly the same.

91
00:04:59,833 --> 00:05:03,400
But with these new average age
and average salary.

92
00:05:04,033 --> 00:05:06,000
Okay,
we're going to check that right away.

93
00:05:06,000 --> 00:05:09,466
Now to do so,
we're going to create a new code cell

94
00:05:09,466 --> 00:05:13,200
where we're going to print
the new matrix of features X.

95
00:05:13,433 --> 00:05:17,700
And let's see if indeed that missing value
that we can clearly see

96
00:05:17,700 --> 00:05:22,200
here as once again
Nan is replaced in this new version of X.

97
00:05:22,200 --> 00:05:25,533
So let's not forget to run these two cells
here,

98
00:05:25,866 --> 00:05:29,400
this first cell
to indeed replace the missing data.

99
00:05:29,400 --> 00:05:33,300
And now this cell
to print the new matrix X.

100
00:05:33,566 --> 00:05:34,800
And there you go.

101
00:05:34,800 --> 00:05:39,433
As we can clearly see that missing salary
in the previous matrix of features

102
00:05:39,433 --> 00:05:44,500
X was indeed replaced
by the average salaries of this column.

103
00:05:44,766 --> 00:05:46,600
Okay. Check that if you want.

104
00:05:46,600 --> 00:05:47,566
But there you go.

105
00:05:47,566 --> 00:05:51,266
Now you have another tool in your data
preprocessing toolkit.

106
00:05:51,500 --> 00:05:52,833
So congratulations.

107
00:05:52,833 --> 00:05:58,633
And now we're going to proceed to a new
tool which is to encode categorical data.
