﻿1
00:00:00,200 --> 00:00:02,633
Hello and welcome to this art tutorial.

2
00:00:02,633 --> 00:00:04,666
Now that we get this curve,
and now that we see

3
00:00:04,666 --> 00:00:07,333
that we have an excellent model
for our problem,

4
00:00:07,333 --> 00:00:09,033
we are going to validate this model.

5
00:00:09,033 --> 00:00:10,066
And this is the model

6
00:00:10,066 --> 00:00:13,700
we were going to choose
to make our final accurate prediction.

7
00:00:13,966 --> 00:00:15,666
So let's do it.

8
00:00:15,666 --> 00:00:19,000
I'm actually going to this new code
section that I prepared for you.

9
00:00:19,000 --> 00:00:22,200
So a first one
to predict the salary associated

10
00:00:22,200 --> 00:00:25,700
to a 6.5 level
according to a linear regression model.

11
00:00:25,800 --> 00:00:29,966
And then the second code section
will predict the salary of the 6.5 level

12
00:00:30,266 --> 00:00:33,266
according to our polynomial regression.

13
00:00:33,266 --> 00:00:36,366
This one right here with a fourth degree.

14
00:00:36,733 --> 00:00:37,000
Okay.

15
00:00:37,000 --> 00:00:40,000
So let's start with the linear regression
prediction.

16
00:00:40,000 --> 00:00:43,100
We're going to do exactly
the same as in simple linear regression.

17
00:00:43,100 --> 00:00:45,300
We're going to use the predict function.

18
00:00:45,300 --> 00:00:49,433
But actually something is going to be
different and will interest many of you.

19
00:00:49,700 --> 00:00:52,266
This time
we're not going to make some predictions

20
00:00:52,266 --> 00:00:55,500
on a whole test
set that is on a vector of observations.

21
00:00:55,800 --> 00:00:58,200
We are going to make a single prediction.

22
00:00:58,200 --> 00:00:59,300
That is, we're going to make

23
00:00:59,300 --> 00:01:03,866
a prediction of a single level,
which is going to be the 6.5 level.

24
00:01:04,200 --> 00:01:05,700
So you're going to see that the syntax

25
00:01:05,700 --> 00:01:09,733
is going to change the technique to make
this prediction is going to change.

26
00:01:09,933 --> 00:01:12,700
But it's actually as easy.

27
00:01:12,700 --> 00:01:13,866
So let's do it.

28
00:01:13,866 --> 00:01:16,833
We're going to call this prediction
widespread as before.

29
00:01:16,833 --> 00:01:19,500
So be careful
why pred was before a vector.

30
00:01:19,500 --> 00:01:22,233
And now it's only going to be
a single prediction value.

31
00:01:22,233 --> 00:01:24,900
So let's still call it why pred anyway.

32
00:01:24,900 --> 00:01:26,533
And same as before.

33
00:01:26,533 --> 00:01:28,900
We're going to take the predict function.

34
00:01:28,900 --> 00:01:31,900
So let's here take predict. Here we go.

35
00:01:32,000 --> 00:01:35,233
And now in this predict function
remember we need to specify first

36
00:01:35,433 --> 00:01:37,533
the regressor
we want to make the predictions with.

37
00:01:37,533 --> 00:01:39,233
So that's the first argument.

38
00:01:39,233 --> 00:01:42,833
And then the second argument
is the new data of which

39
00:01:42,833 --> 00:01:44,900
we want to make the predictions.

40
00:01:44,900 --> 00:01:47,066
Okay. So let's input the first argument.

41
00:01:47,066 --> 00:01:48,566
The first argument is our regressor.

42
00:01:48,566 --> 00:01:53,000
And since we're predicting the result
with the linear regression regressor.

43
00:01:53,000 --> 00:01:56,100
And since we called this regressor linear,

44
00:01:56,866 --> 00:02:00,066
then the first argument
we need to input here is Lin rec.

45
00:02:01,666 --> 00:02:02,033
All right.

46
00:02:02,033 --> 00:02:03,800
So that's our regressor.

47
00:02:03,800 --> 00:02:08,566
And now we need to input the new data 
of which we want to make the prediction.

48
00:02:08,766 --> 00:02:14,600
So this new data is as you understood, a
single element a single observation point.

49
00:02:15,033 --> 00:02:19,600
And this observation
point is actually the 6.5 level here.

50
00:02:19,600 --> 00:02:23,500
So what we'll do now is to actually,
you know,

51
00:02:23,500 --> 00:02:27,333
since this 6.5 level
doesn't exist in our data set,

52
00:02:27,666 --> 00:02:32,366
we actually need to create a new data
frame containing the 6.5 value.

53
00:02:32,400 --> 00:02:35,933
We're not going to add
this 6.5 level in our data set.

54
00:02:36,166 --> 00:02:40,800
We're going to create a new data
set of only one line and only one column.

55
00:02:41,100 --> 00:02:43,166
That is of only one cell actually.

56
00:02:43,166 --> 00:02:46,633
And this cell will contain the 6.5 level.

57
00:02:46,933 --> 00:02:47,933
So let's do this.

58
00:02:47,933 --> 00:02:50,566
The syntax in order to do
that is very simple.

59
00:02:50,566 --> 00:02:54,000
We just need to type here
dataframe separated by a dot.

60
00:02:54,433 --> 00:02:57,366
So frame here we go data frame.

61
00:02:57,366 --> 00:03:00,633
And now as you can see it
automatically added some parenthesis.

62
00:03:01,000 --> 00:03:04,633
And in these parenthesis
we need to input level

63
00:03:06,500 --> 00:03:09,500
equals 6.5.

64
00:03:09,533 --> 00:03:10,733
And that's it.

65
00:03:10,733 --> 00:03:14,000
That's actually all you need to do
to make a single prediction

66
00:03:14,000 --> 00:03:16,800
using the predict function
with a specific regressor.

67
00:03:16,800 --> 00:03:18,900
That is here the linear regressor.

68
00:03:18,900 --> 00:03:20,000
Let's check it out.

69
00:03:20,000 --> 00:03:24,333
Let's select this line and execute okay.

70
00:03:24,333 --> 00:03:26,600
Why pred correctly generated.

71
00:03:26,600 --> 00:03:30,000
So as you can see why pred is the values
here, we can already see

72
00:03:30,166 --> 00:03:33,433
the predicted value of the 6.5
level salary

73
00:03:33,966 --> 00:03:37,100
and it's actually $330,000,

74
00:03:37,500 --> 00:03:41,366
so much higher than what
this employee mentioned its salary was.

75
00:03:41,366 --> 00:03:43,500
So that's actually a good start for us.

76
00:03:43,500 --> 00:03:46,300
But remember, we don't want to keep
the linear regression model.

77
00:03:46,300 --> 00:03:49,666
We want to keep the most accurate
regression model.

78
00:03:49,933 --> 00:03:52,100
That is of course
the polynomial regression model.

79
00:03:52,100 --> 00:03:55,333
So what we're going to do now
is make this same prediction.

80
00:03:55,566 --> 00:03:58,633
But this time according to our
polynomial regression model.
