﻿1
00:00:00,266 --> 00:00:01,433
All right, let's do this.

2
00:00:01,433 --> 00:00:04,100
Let's predict the test set results.

3
00:00:04,100 --> 00:00:08,266
So as I said in the previous tutorial,
we're going to display two vectors.

4
00:00:08,433 --> 00:00:12,033
First one being the vector
of the real profits in the test set,

5
00:00:12,300 --> 00:00:16,200
and the second one being the vector of the
predicted profits in the same test set.

6
00:00:16,466 --> 00:00:18,300
And we're going to compare next
to each other

7
00:00:18,300 --> 00:00:21,633
to see if our predictions
are close to the real results.

8
00:00:22,033 --> 00:00:23,100
All right, let's do this.

9
00:00:23,100 --> 00:00:26,100
Let's start by creating a new code cell.

10
00:00:26,133 --> 00:00:30,033
So the first thing I asked you to do was
to indeed get that vector of predictions.

11
00:00:30,200 --> 00:00:34,033
So first, since we want to get a vector,
I'm going to introduce a new variable,

12
00:00:34,200 --> 00:00:37,200
which I'm going to call as usual y pred.

13
00:00:37,633 --> 00:00:41,600
So that's going to be the vector
of the predicted profits in the test set.

14
00:00:42,233 --> 00:00:44,300
And now how do we get our predictions.

15
00:00:44,300 --> 00:00:45,600
Well you know the solution.

16
00:00:45,600 --> 00:00:49,200
It's actually exactly
the same as with simple linear regression.

17
00:00:49,466 --> 00:00:52,766
We first need to take our regressor
object.

18
00:00:52,833 --> 00:00:56,566
You know our multiple linear
regression model from which

19
00:00:56,566 --> 00:01:00,000
we're going to call that predict method.

20
00:01:00,566 --> 00:01:05,866
And of course in this method we have to
input the features of the test set.

21
00:01:05,866 --> 00:01:09,866
You know exactly these ones
not including the prophet of course.

22
00:01:10,100 --> 00:01:13,800
These are all the features with first
the encoded variables

23
00:01:13,800 --> 00:01:17,700
for the state, then the R&D spend
and the administration spend.

24
00:01:17,900 --> 00:01:19,666
And the marketing spend.

25
00:01:19,666 --> 00:01:20,000
All right.

26
00:01:20,000 --> 00:01:24,300
So these are all the features
which we need to input in our predict

27
00:01:24,300 --> 00:01:27,733
method here
in order to predict the profits.

28
00:01:27,766 --> 00:01:30,766
So let's do this X test.

29
00:01:31,000 --> 00:01:31,866
And there you go.

30
00:01:31,866 --> 00:01:34,866
Now we get our vector of predicted
profits.

31
00:01:35,200 --> 00:01:37,233
Perfect. And now the next step.

32
00:01:37,233 --> 00:01:38,233
The thing that I'm going to do

33
00:01:38,233 --> 00:01:42,666
is I'm going to call numpy,
which has a shortcut NP from which I'm

34
00:01:42,666 --> 00:01:48,066
going to call one of its function
which is set underscore print options.

35
00:01:48,633 --> 00:01:52,033
And now in the parentheses
I'm just going to enter

36
00:01:52,033 --> 00:01:55,033
precision equals to.

37
00:01:55,033 --> 00:01:55,433
All right.

38
00:01:55,433 --> 00:02:00,633
And this will display any numerical value
with only two decimals after comma.

39
00:02:00,833 --> 00:02:01,166
All right.

40
00:02:01,166 --> 00:02:02,766
So this will be much more beautiful

41
00:02:02,766 --> 00:02:06,000
to visualize
and mostly much easier to visualize.

42
00:02:06,600 --> 00:02:07,700
Okay. Good.

43
00:02:07,700 --> 00:02:08,966
And now final step.

44
00:02:08,966 --> 00:02:12,233
This is a step where we're going
to display the two vectors

45
00:02:12,233 --> 00:02:15,933
of the real profits and of the predicted
profits together next to each other.

46
00:02:16,433 --> 00:02:20,366
And to do this well we're going to use,
you know, a classic trick with numpy.

47
00:02:20,400 --> 00:02:21,466
Concatenate.

48
00:02:21,466 --> 00:02:25,533
Concatenate is a function of numpy
that allows to concatenate

49
00:02:25,533 --> 00:02:29,300
either vertically or horizontally
two vectors or even arrays.

50
00:02:29,600 --> 00:02:33,166
And so now we're going to use
the concatenate function to concatenate

51
00:02:33,166 --> 00:02:37,133
vertically are two vectors of the
real profits and the predicted profits.

52
00:02:37,433 --> 00:02:39,666
All right
so let's do this. Just follow me.

53
00:02:39,666 --> 00:02:42,133
I will explain everything as I'm coding.

54
00:02:42,133 --> 00:02:44,700
So first we're simply going to start
with a print.

55
00:02:44,700 --> 00:02:48,300
You know because we want to print
that concatenation of the two vectors.

56
00:02:48,666 --> 00:02:49,700
Then here we go.

57
00:02:49,700 --> 00:02:54,133
We're going to call that concatenate
function by numpy from which there we go.

58
00:02:54,133 --> 00:02:58,533
We call the concatenate function
perfect parentheses.

59
00:02:59,166 --> 00:03:02,166
And now be careful
because this is always a bit confusing.

60
00:03:02,333 --> 00:03:07,200
The concatenate function by numpy
actually expects as a first argument.

61
00:03:07,266 --> 00:03:11,733
Check the cell here, the couple of arrays
you want to concatenate

62
00:03:11,733 --> 00:03:13,300
or you know, vectors.

63
00:03:13,300 --> 00:03:15,366
So actually all the description is here.

64
00:03:15,366 --> 00:03:19,700
You know A1, a2 is a sequence of arrays
you want to concatenate.

65
00:03:19,700 --> 00:03:21,266
And that must have the same shape.

66
00:03:21,266 --> 00:03:25,066
Well that's perfect for us because
of course our vector of predicted profits

67
00:03:25,066 --> 00:03:28,066
and the vector of real profits
have exactly the same shape, meaning

68
00:03:28,133 --> 00:03:32,333
they are one dimensional vector
containing the same number of profits.

69
00:03:32,433 --> 00:03:33,766
Okay, so that's good.

70
00:03:33,766 --> 00:03:37,600
But these vectors that we want to
concatenate have to be

71
00:03:37,866 --> 00:03:39,566
in, you know, some parentheses.

72
00:03:39,566 --> 00:03:42,000
And these are actually the first argument.

73
00:03:42,000 --> 00:03:45,866
You know, this double of arrays of vectors
is the first argument

74
00:03:45,866 --> 00:03:47,066
input in parentheses.

75
00:03:47,066 --> 00:03:49,733
So here I'm going to add new parentheses.

76
00:03:49,733 --> 00:03:52,166
All right. This is the first argument.

77
00:03:52,166 --> 00:03:54,066
And what is going to be inside
this parentheses.

78
00:03:54,066 --> 00:03:56,033
Well of course are two vectors

79
00:03:56,033 --> 00:03:59,033
the vector of predicted profits
and the vector of real profits.

80
00:03:59,700 --> 00:04:00,000
All right.

81
00:04:00,000 --> 00:04:03,000
So let's first add
the vector of predicted profit.

82
00:04:03,066 --> 00:04:06,066
This is of course what pred.

83
00:04:06,233 --> 00:04:09,666
And now because we want to display them
vertically

84
00:04:09,666 --> 00:04:11,533
and not horizontally remember that.

85
00:04:11,533 --> 00:04:14,700
You know actually I can show this to you
in data preprocessing tools.

86
00:04:14,700 --> 00:04:17,700
We actually printed Y at some point.

87
00:04:18,066 --> 00:04:18,700
There we go.

88
00:04:18,700 --> 00:04:20,933
You know,
remember when we print the dependent

89
00:04:20,933 --> 00:04:24,200
variable vector, since it is vector
it is displayed horizontally.

90
00:04:24,600 --> 00:04:26,400
I actually prefer to display,

91
00:04:26,400 --> 00:04:29,966
you know, our two vectors of predictions
in real profit vertically.

92
00:04:30,100 --> 00:04:34,200
So now I'm going to add another trick
to put that vertically.

93
00:04:34,200 --> 00:04:36,766
You know, to convert that
from being horizontal to vertical.

94
00:04:36,766 --> 00:04:40,733
And the trick to do that is just add
dot reshape.

95
00:04:41,066 --> 00:04:43,100
Reshape is an attribute function
that allows

96
00:04:43,100 --> 00:04:45,766
to, you know,
reshape your vectors or arrays.

97
00:04:45,766 --> 00:04:49,100
And in order to reshape a vector
from being horizontal to vertical,

98
00:04:49,233 --> 00:04:52,233
well, we just need to add
as input of this function.

99
00:04:52,400 --> 00:04:55,400
First, the number of elements in y print.

100
00:04:55,466 --> 00:04:57,400
And that's actually so far
the number of columns

101
00:04:57,400 --> 00:04:59,033
because you know it is horizontal.

102
00:04:59,033 --> 00:05:01,866
And to get that number
we can simply use the

103
00:05:01,866 --> 00:05:05,200
Len function,
which returns the length of a vector.

104
00:05:05,500 --> 00:05:09,766
And therefore here
I'm going to input y Brett okay.

105
00:05:10,200 --> 00:05:13,233
So this is the first element
of the reshape function.

106
00:05:13,233 --> 00:05:15,933
And the second one is just one.

107
00:05:15,933 --> 00:05:17,033
All right.

108
00:05:17,033 --> 00:05:17,966
So what does it mean.

109
00:05:17,966 --> 00:05:22,133
That means that you want to reshape
your y vector into an array.

110
00:05:22,166 --> 00:05:24,933
Having Len y red rows.

111
00:05:24,933 --> 00:05:28,300
Meaning you know that the number of rows
will be equal to the number of strips.

112
00:05:28,633 --> 00:05:30,666
And then just one column.

113
00:05:30,666 --> 00:05:31,100
All right.

114
00:05:31,100 --> 00:05:32,000
That's what it means.

115
00:05:32,000 --> 00:05:34,500
So it's good that you know this
reshape trick.

116
00:05:34,500 --> 00:05:36,133
Now you know how to reshape your vectors.

117
00:05:36,133 --> 00:05:38,500
And you will see that
it is actually going to be much nicer.
