﻿1
00:00:00,166 --> 00:00:02,766
Hello and welcome to this art tutorial.

2
00:00:02,766 --> 00:00:05,100
So we did the big part
the cleaning process.

3
00:00:05,100 --> 00:00:07,500
We started by creating this corpus here.

4
00:00:07,500 --> 00:00:09,633
Then we put all the words in lower cases.

5
00:00:09,633 --> 00:00:12,633
Then we removed all the numbers,
all the punctuation

6
00:00:12,866 --> 00:00:15,666
and we also removed
all the non relevant words.

7
00:00:15,666 --> 00:00:19,800
And finally we took the roots
of all the words in the 1000 reviews.

8
00:00:20,466 --> 00:00:24,033
So all these steps considerably simplified
the words in the reviews.

9
00:00:24,033 --> 00:00:27,166
So thanks to all these steps
here, all the simplifications.

10
00:00:27,400 --> 00:00:28,766
Well, our final sparse

11
00:00:28,766 --> 00:00:32,066
matrix of features
will get a much smaller number of columns.

12
00:00:32,300 --> 00:00:35,300
So that's good for algorithm
because we reduced sparsity.

13
00:00:35,700 --> 00:00:38,400
And now we have one
final little step to do

14
00:00:38,400 --> 00:00:41,700
that will not be done to simplify
even more of the corpus.

15
00:00:42,000 --> 00:00:44,900
But that just consists
of removing the extra spaces.

16
00:00:44,900 --> 00:00:48,300
Because by doing all these simplifications
here, well,

17
00:00:48,300 --> 00:00:52,133
you know, we removed some things,
and those things that were removed

18
00:00:52,333 --> 00:00:55,200
could actually have been replaced
by a space,

19
00:00:55,200 --> 00:00:58,333
and that would cause an extra space
in the review.

20
00:00:58,566 --> 00:00:59,933
And therefore, if we want to have

21
00:00:59,933 --> 00:01:03,566
perfectly clean reviews,
we must remove these extra spaces.

22
00:01:04,033 --> 00:01:08,300
If we remove all the extra spaces, well,
the columns in our final sparse matrix

23
00:01:08,300 --> 00:01:11,466
and features will only contain the words
that are relevant

24
00:01:11,700 --> 00:01:14,400
and will not contain any space
or anything else.

25
00:01:14,400 --> 00:01:18,200
So that's what we'll remove in this final
step of the cleaning process.

26
00:01:18,366 --> 00:01:20,200
So let's do it again.

27
00:01:20,200 --> 00:01:21,533
It's going to be very simple.

28
00:01:21,533 --> 00:01:24,100
Will copy this line here.

29
00:01:24,100 --> 00:01:27,366
Copy and paste it below and replace

30
00:01:27,366 --> 00:01:31,600
stem document here by strip whitespace.

31
00:01:31,600 --> 00:01:37,300
Here it is pressing enter and all good
final step ready to be executed.

32
00:01:37,300 --> 00:01:41,066
And therefore the whole cleaning process
is ready to be completed.

33
00:01:41,600 --> 00:01:44,400
So before we execute this line,
let's have a look

34
00:01:44,400 --> 00:01:47,400
at a review that now has an extra space.

35
00:01:47,400 --> 00:01:51,800
I remember that when we removed the number
in the review 841.

36
00:01:52,033 --> 00:01:55,000
Well, we got an extra space. Here it is.

37
00:01:55,000 --> 00:01:58,833
You know,
before the review was for 40 bucks ahead.

38
00:01:59,100 --> 00:02:00,400
I really expect better food.

39
00:02:00,400 --> 00:02:04,466
And when we applied the remove numbers
here to the corpus, well,

40
00:02:04,466 --> 00:02:06,333
the number 40 here disappeared.

41
00:02:06,333 --> 00:02:07,900
But it actually didn't disappear.

42
00:02:07,900 --> 00:02:10,900
Was just replaced
by this extra space here.

43
00:02:11,000 --> 00:02:15,033
Because indeed we can see that
we have two spaces between four and books.

44
00:02:15,366 --> 00:02:17,966
And what we want to get
is only one space here.

45
00:02:17,966 --> 00:02:20,433
So we're
just removing the extra space here.

46
00:02:20,433 --> 00:02:21,633
So let's check it out.

47
00:02:21,633 --> 00:02:26,400
When we select this final line of code
of our cleaning process.

48
00:02:26,600 --> 00:02:28,733
Well,
let's make sure that this extra space

49
00:02:28,733 --> 00:02:32,166
in the review 841
disappears, this time for good.

50
00:02:32,700 --> 00:02:33,066
All right.

51
00:02:33,066 --> 00:02:35,800
So I'm going to press Command
plus enter to execute.

52
00:02:35,800 --> 00:02:36,633
Here we go.

53
00:02:36,633 --> 00:02:40,033
And now let's have a look
at the review 841.

54
00:02:40,033 --> 00:02:42,900
So let's press
the up arrow here to find it.

55
00:02:44,500 --> 00:02:45,433
Here it is.

56
00:02:45,433 --> 00:02:48,366
We're going to get the new version
of the review 841

57
00:02:48,366 --> 00:02:51,966
after we applied
this strict whitespace to the corpus.

58
00:02:52,166 --> 00:02:54,866
So let's do it pressing command here.

59
00:02:54,866 --> 00:02:56,766
Well not only the space was removed,

60
00:02:56,766 --> 00:02:59,266
we can see that
we don't have any extra space here.

61
00:02:59,266 --> 00:03:02,366
But also we get all the other steps
of the cleaning process.

62
00:03:02,666 --> 00:03:06,433
Since indeed we can clearly see that
the non relevant words were removed.

63
00:03:06,666 --> 00:03:09,866
So the non relevant words
were for example four.

64
00:03:10,066 --> 00:03:11,633
So this was removed.

65
00:03:11,633 --> 00:03:15,366
And then a I, I think that's all.

66
00:03:15,966 --> 00:03:20,533
Yes. As you can see a was removed here
and I was removed here.

67
00:03:21,066 --> 00:03:23,366
Okay. So that's the first thing
we see very clearly.

68
00:03:23,366 --> 00:03:26,800
And the second thing we see very clearly
is the stemming of course,

69
00:03:26,800 --> 00:03:29,700
because we hardly recognize the words
here.

70
00:03:29,700 --> 00:03:31,933
Box was replaced by bug.

71
00:03:31,933 --> 00:03:34,800
You know, it's
not only about the past tense of verbs,

72
00:03:34,800 --> 00:03:37,900
it's also about the singular
and plural of nouns.

73
00:03:38,133 --> 00:03:41,100
So books became bug

74
00:03:41,100 --> 00:03:45,900
and head was not replaced
because the root of head is head.

75
00:03:45,900 --> 00:03:48,900
So that's why we kept head here.

76
00:03:49,166 --> 00:03:51,600
And then, really,

77
00:03:51,600 --> 00:03:54,600
really with a
why became really with an eye.

78
00:03:54,766 --> 00:03:57,466
So that's
how are simply interpreted the root.

79
00:03:57,466 --> 00:03:59,233
So that's okay. It's not a mistake.

80
00:03:59,233 --> 00:04:00,800
And finally expect better.

81
00:04:00,800 --> 00:04:04,966
And food were not replaced
because you know these are kind of already

82
00:04:04,966 --> 00:04:06,133
the root of the words.

83
00:04:06,133 --> 00:04:09,133
And we cannot really simplify
the words more.

84
00:04:09,200 --> 00:04:09,566
All right.

85
00:04:09,566 --> 00:04:10,833
So that's a good example.

86
00:04:10,833 --> 00:04:12,600
We can clearly see what happened here.

87
00:04:12,600 --> 00:04:16,033
And for what we just did here
with the stripped whitespace

88
00:04:16,166 --> 00:04:18,900
we can see that
this extra space here was removed.

89
00:04:18,900 --> 00:04:20,733
Well there is still a space here.

90
00:04:20,733 --> 00:04:25,200
But you know before there was the four
and two spaces between four and book.

91
00:04:25,466 --> 00:04:28,400
And then after the stopwords step,
the four was removed.

92
00:04:28,400 --> 00:04:32,600
So we can still see a space here
before bug, but that's actually one space

93
00:04:32,600 --> 00:04:33,600
instead of two.

94
00:04:33,600 --> 00:04:34,300
If we want to get

95
00:04:34,300 --> 00:04:38,033
even more convinced, well, we can see that
we have an extra space here

96
00:04:38,300 --> 00:04:39,433
between love and place.

97
00:04:39,433 --> 00:04:42,500
This review is the first review
in the previous version

98
00:04:42,500 --> 00:04:45,900
of the corpus, before we applied the strip
whitespace here.

99
00:04:46,200 --> 00:04:48,766
And so now we'll see that
if we have a look at the first review

100
00:04:48,766 --> 00:04:54,000
of the new version of our corpus, well,
this extra space here will be removed,

101
00:04:54,200 --> 00:04:57,666
and we will get only one space
between love and place instead of two.

102
00:04:57,666 --> 00:04:59,733
Here. Let's check it out.

103
00:04:59,733 --> 00:05:00,266
Here we go.

104
00:05:00,266 --> 00:05:04,533
We can clearly see that we have only one
space here instead of two spaces here.

105
00:05:04,900 --> 00:05:06,033
All right, so good.

106
00:05:06,033 --> 00:05:08,566
Removing the extra spaces worked properly.

107
00:05:08,566 --> 00:05:11,466
And so we are done
with the cleaning process.

108
00:05:11,466 --> 00:05:12,366
So great.

109
00:05:12,366 --> 00:05:16,566
That means that now we are ready to build
the sparse matrix of features.

110
00:05:16,833 --> 00:05:18,566
We'll do that in the next tutorial.

111
00:05:18,566 --> 00:05:20,400
And until then enjoy machine learning.
