﻿1
00:00:00,133 --> 00:00:02,066
All right, here we go for that final step.

2
00:00:02,066 --> 00:00:05,966
Predicting the salary of the position
level 6.5

3
00:00:05,966 --> 00:00:09,333
with both linear
regression and polynomial regression.

4
00:00:09,333 --> 00:00:10,533
So we're going to start with

5
00:00:10,533 --> 00:00:14,166
linear regression first
therefore creating a new code cell here.

6
00:00:14,500 --> 00:00:15,800
Let's do a step by step.

7
00:00:15,800 --> 00:00:19,700
Of course the first step you had to do
here was to take your linear

8
00:00:19,700 --> 00:00:23,433
regressor object, you know,
from the simple linear regression model.

9
00:00:23,700 --> 00:00:26,800
And this object was named Lin Regge.

10
00:00:27,300 --> 00:00:28,766
This one.

11
00:00:28,766 --> 00:00:31,166
And then very simply from this object

12
00:00:31,166 --> 00:00:34,166
you're going to call the predict method

13
00:00:34,500 --> 00:00:38,933
to predict the salary of the position
level 6.5.

14
00:00:39,700 --> 00:00:40,166
All right.

15
00:00:40,166 --> 00:00:41,966
So what exactly do you need

16
00:00:41,966 --> 00:00:45,566
to input here in this brick
method to get such a predicted salary?

17
00:00:45,833 --> 00:00:50,366
Well I'm sure some of you
try to directly input 6.5,

18
00:00:50,666 --> 00:00:54,300
but unfortunately this is not that simple
because indeed

19
00:00:54,300 --> 00:00:59,700
you always have to input your observations
in an array, right?

20
00:00:59,700 --> 00:01:02,666
That's
what is specified here in the parameters.

21
00:01:02,666 --> 00:01:06,666
The input of the brick method should be
an array like or a sparse matrix.

22
00:01:06,900 --> 00:01:10,533
So an array like this for example
a numpy array or a simple array

23
00:01:10,533 --> 00:01:12,666
with double pair of square brackets.

24
00:01:12,666 --> 00:01:14,266
Well you know I said it.

25
00:01:14,266 --> 00:01:17,133
That's
exactly how you need to input an array.

26
00:01:17,133 --> 00:01:22,333
Because generally in Python an array
is built with pairs of square brackets.

27
00:01:22,700 --> 00:01:25,500
If you add just a single pair

28
00:01:25,500 --> 00:01:29,166
of square brackets here,
well that actually creates a list.

29
00:01:29,400 --> 00:01:32,033
Or you can also see it as a vector.

30
00:01:32,033 --> 00:01:35,200
But in order to create an array, well,
you know, an array contains

31
00:01:35,366 --> 00:01:38,200
several dimensions
like two dimensions in our case here.

32
00:01:38,200 --> 00:01:41,633
And therefore you need to add
a double pair of square brackets.

33
00:01:42,066 --> 00:01:44,633
What does this
double pair of square brackets mean?

34
00:01:44,633 --> 00:01:49,500
Well, the first pair of square brackets
here corresponds to the first dimension,

35
00:01:49,733 --> 00:01:53,633
and the second pair of square brackets
here corresponds to the second dimension.

36
00:01:53,633 --> 00:01:57,566
So the first dimension is actually
corresponding to the rows in your array,

37
00:01:57,766 --> 00:02:00,833
and the second dimension
is corresponding to your column.

38
00:02:00,933 --> 00:02:04,800
So for example, if I add just to show you,
if I add for example five here,

39
00:02:05,066 --> 00:02:09,433
that would create actually
an array of one row and two columns.

40
00:02:09,733 --> 00:02:14,133
And if I add here a comma
and then another pair of square brackets,

41
00:02:14,133 --> 00:02:17,333
and then for example two and three,
well this would

42
00:02:17,333 --> 00:02:20,533
create indeed an array of two rows
and two columns.

43
00:02:20,700 --> 00:02:23,233
In the first row
you would have 6.5 and five.

44
00:02:23,233 --> 00:02:25,666
And in second row
you would have two and three.

45
00:02:25,666 --> 00:02:28,666
So you see the first square bracket here
corresponds

46
00:02:28,666 --> 00:02:31,666
to the rows and the second one
corresponds to the columns.

47
00:02:31,766 --> 00:02:32,033
All right.

48
00:02:32,033 --> 00:02:35,200
So let's
go back to where we were exactly this.

49
00:02:35,400 --> 00:02:39,066
So this indeed is exactly
the format expected by this break

50
00:02:39,133 --> 00:02:41,433
method an array of two dimensions.

51
00:02:41,433 --> 00:02:44,933
Even if this only contains one cell,
you know one value.

52
00:02:44,933 --> 00:02:46,733
Well, it has to be in this format.

53
00:02:46,733 --> 00:02:49,800
And now while you're ready
to get that prediction.

54
00:02:49,800 --> 00:02:53,066
So before we execute this,
let's remember that

55
00:02:53,066 --> 00:02:57,733
this person asked for a 160 K salary.

56
00:02:57,733 --> 00:03:01,833
And this is justified
by the fact that this person earned a 160

57
00:03:01,833 --> 00:03:05,933
K salary in the previous company,
which is what we have to check right now.

58
00:03:05,933 --> 00:03:09,933
And we're going to check that first
with the linear regression model,

59
00:03:09,933 --> 00:03:15,300
which returns
indeed a salary of $330,000 per year.

60
00:03:15,733 --> 00:03:20,766
So if we actually use this model
to negotiate with that person,

61
00:03:21,033 --> 00:03:23,233
we would actually find it
super weird, right?

62
00:03:23,233 --> 00:03:26,333
Because this predicted salary is way over

63
00:03:26,333 --> 00:03:30,000
the real salary that this person
had in its previous company.

64
00:03:30,333 --> 00:03:32,666
So clearly the prediction here is wrong.

65
00:03:32,666 --> 00:03:35,400
And, you know, that's what
we can clearly see on the graph.

66
00:03:35,400 --> 00:03:38,266
You know, the graph
above showing the results of the linear

67
00:03:38,266 --> 00:03:42,233
regression
model 6.5 is somewhere around here.

68
00:03:42,466 --> 00:03:44,100
And indeed,
if we want to get the prediction,

69
00:03:44,100 --> 00:03:48,200
we have to project it to that blue
regression line and then projected

70
00:03:48,200 --> 00:03:51,266
again to the vertical axis,
which is around here.

71
00:03:51,266 --> 00:03:52,800
And indeed that's what we get.

72
00:03:52,800 --> 00:03:55,400
You know, this is multiplied
by ten to the power of six.

73
00:03:55,400 --> 00:03:58,833
So indeed we get something around 330.
