1
00:00:00,300 --> 00:00:01,700
So here we are in.

2
00:00:01,700 --> 00:00:04,366
And now that you understand
feature scaling

3
00:00:04,366 --> 00:00:07,166
let's apply it to the training set.

4
00:00:07,166 --> 00:00:12,300
And the test set here we will just write
two lines of code which are.

5
00:00:12,300 --> 00:00:15,300
So the first line is training sets.

6
00:00:15,766 --> 00:00:17,833
So two training sets already exists right.

7
00:00:17,833 --> 00:00:19,800
It's this one. The training set.

8
00:00:19,800 --> 00:00:21,633
As you can see it's not scaled.

9
00:00:21,633 --> 00:00:23,200
These are

10
00:00:23,200 --> 00:00:25,733
it contains the raw values.

11
00:00:25,733 --> 00:00:29,066
So training set equals and simply.

12
00:00:29,066 --> 00:00:32,066
Now we're going to write scale.

13
00:00:35,400 --> 00:00:37,733
And then training set.

14
00:00:37,733 --> 00:00:39,333
That's all that scales.

15
00:00:39,333 --> 00:00:41,600
Your training set is in a grid.

16
00:00:41,600 --> 00:00:42,766
And same for the test set.

17
00:00:42,766 --> 00:00:45,766
We will copy this line.

18
00:00:46,000 --> 00:00:47,233
Paste it here.

19
00:00:47,233 --> 00:00:50,233
And we will change training into test.

20
00:00:50,900 --> 00:00:53,900
And here as well.

21
00:00:54,866 --> 00:00:56,533
Okay so here I just wrote this.

22
00:00:56,533 --> 00:00:58,766
There is something important
to understand.

23
00:00:58,766 --> 00:01:00,000
This will be the feature

24
00:01:00,000 --> 00:01:03,000
scaling block of code
that we will be using in our template.

25
00:01:03,166 --> 00:01:07,933
However let's see what happens
when I select this code and execute it.

26
00:01:08,800 --> 00:01:11,000
So let's select this execute

27
00:01:12,000 --> 00:01:12,833
okay.

28
00:01:12,833 --> 00:01:15,833
As you can see here I obtained two errors
Eren.

29
00:01:15,833 --> 00:01:19,666
Code means x must be numeric
for the training set, and the test set.

30
00:01:20,566 --> 00:01:22,800
Can you guess what the problem is?

31
00:01:22,800 --> 00:01:24,000
The problem is that.

32
00:01:24,000 --> 00:01:26,400
Okay, well,
it tells us what the problem is.

33
00:01:26,400 --> 00:01:28,766
It tells us that X must be numeric.

34
00:01:28,766 --> 00:01:30,700
But what does that mean?

35
00:01:30,700 --> 00:01:32,066
Well, first what is this x?

36
00:01:32,066 --> 00:01:34,833
X is for this line of code,
the training set.

37
00:01:34,833 --> 00:01:36,700
And for this line of code the test set.

38
00:01:36,700 --> 00:01:38,433
So let's forget the test set for a second.

39
00:01:38,433 --> 00:01:40,266
And let's focus on the training set.

40
00:01:40,266 --> 00:01:41,700
So x is the training set.

41
00:01:41,700 --> 00:01:44,700
And it says that the training set
must be numeric.

42
00:01:44,700 --> 00:01:47,533
So let's look at our training set okay.

43
00:01:47,533 --> 00:01:49,300
Well the training set looks numeric.

44
00:01:49,300 --> 00:01:53,000
Right here we have numeric values numeric
values numeric values numeric values.

45
00:01:53,533 --> 00:01:57,433
But no actually there are two columns
that don't have numeric values.

46
00:01:58,100 --> 00:02:01,633
It's this one the country
and this one purchased.

47
00:02:02,200 --> 00:02:03,933
And you remember why?

48
00:02:03,933 --> 00:02:06,966
Well, it's because
before we had the country written in text

49
00:02:07,233 --> 00:02:10,233
and the purchase column written in text
with yes or no,

50
00:02:10,233 --> 00:02:14,566
and we changed that
by putting the categories as factors.

51
00:02:15,000 --> 00:02:16,600
That's what we did here.

52
00:02:16,600 --> 00:02:21,000
As you remember,
data set country equals factor

53
00:02:21,300 --> 00:02:24,933
of the different levels and labels
and a factor.

54
00:02:24,933 --> 00:02:27,866
And R is not a numeric number.

55
00:02:27,866 --> 00:02:31,300
And when you apply the scale here
x must be numeric.

56
00:02:31,300 --> 00:02:36,300
That means that all the columns in x
that is the training set must be numeric.

57
00:02:37,633 --> 00:02:39,133
So this time we are going to

58
00:02:39,133 --> 00:02:42,733
exclude the categories
from the feature scaling.

59
00:02:42,966 --> 00:02:45,966
We're not going to apply feature
scaling on those columns.

60
00:02:46,133 --> 00:02:47,366
So that's very simple.

61
00:02:47,366 --> 00:02:51,800
All we need to do is to take the columns
we're interested in

62
00:02:52,500 --> 00:02:56,800
which are well the indexes of the column
we want to scale which are indexes.

63
00:02:57,200 --> 00:02:59,066
So indexes in R start at one.

64
00:02:59,066 --> 00:03:01,566
So that's 123.

65
00:03:01,566 --> 00:03:04,233
So that's the second and third index

66
00:03:04,233 --> 00:03:07,633
we want to take to scale the age
and the salary columns.

67
00:03:08,066 --> 00:03:10,966
So two and three let's input it.

68
00:03:10,966 --> 00:03:14,200
We need to specify here two column three.

69
00:03:14,466 --> 00:03:16,266
And that gets what we want.

70
00:03:16,266 --> 00:03:19,266
So now I'm going to copy this copy

71
00:03:20,100 --> 00:03:22,833
paste it here.

72
00:03:22,833 --> 00:03:25,366
Here.

73
00:03:25,366 --> 00:03:26,400
And here.

74
00:03:26,400 --> 00:03:27,900
All right. And now it's ready.

75
00:03:27,900 --> 00:03:30,400
Let's have a look at the training set
and the test sets.

76
00:03:30,400 --> 00:03:33,000
Not scaled not scaled.

77
00:03:33,000 --> 00:03:34,466
Let's go back to our code.

78
00:03:34,466 --> 00:03:36,066
Let's select this.

79
00:03:36,066 --> 00:03:39,066
And now we shouldn't get an error.

80
00:03:39,866 --> 00:03:41,466
Come in a control percentage to execute.

81
00:03:42,600 --> 00:03:43,366
Here we go.

82
00:03:43,366 --> 00:03:44,933
It executed properly.

83
00:03:44,933 --> 00:03:47,800
And now let's look at the training set.

84
00:03:47,800 --> 00:03:49,800
All scaled properly. Perfect.

85
00:03:49,800 --> 00:03:51,866
And the test set

86
00:03:51,866 --> 00:03:54,033
all scaled properly perfect.

87
00:03:54,033 --> 00:03:57,300
And now our data is ready
to offer a good precision

88
00:03:57,300 --> 00:04:00,833
and good accuracy and a fast
work of the machine learning models.

89
00:04:00,833 --> 00:04:05,100
And by that I mean that the machine
learning models will converge rapidly.

90
00:04:06,866 --> 00:04:08,600
Okay, so that's it for feature scaling.

91
00:04:08,600 --> 00:04:10,733
Now you know how to apply features.

92
00:04:10,733 --> 00:04:13,833
Getting to your data in Python
in R congratulations.

93
00:04:13,833 --> 00:04:17,066
And mostly congratulations because we did

94
00:04:17,066 --> 00:04:20,366
all the required steps
to preprocess our data.

95
00:04:20,366 --> 00:04:21,733
Feature scaling was the last one

96
00:04:21,733 --> 00:04:25,766
because the next tutorial will be
about this data pre-processing template,

97
00:04:25,833 --> 00:04:29,666
and I will just explain how we are going
to use it in our machine learning models.

98
00:04:29,666 --> 00:04:33,700
It's going to be very fast and practical,
so we are done with the data

99
00:04:33,700 --> 00:04:34,366
pre-processing.

100
00:04:34,366 --> 00:04:37,000
Congratulations,
you did the most difficult part

101
00:04:37,000 --> 00:04:38,700
and now it's time to have fun.

102
00:04:38,700 --> 00:04:42,733
It's time to start making the models
and I can't wait to start them with you.

103
00:04:43,233 --> 00:04:45,300
So thank you for watching this tutorial.

104
00:04:45,300 --> 00:04:47,100
I look forward
to seeing you on the next one.

105
00:04:47,100 --> 00:04:48,966
And until then, enjoy machine learning.