﻿1
00:00:00,200 --> 00:00:01,066
Okay my friends.

2
00:00:01,066 --> 00:00:04,500
So we just covered the data
preprocessing part and we covered it

3
00:00:04,500 --> 00:00:08,100
in a flash light
thanks to our data preprocessing template.

4
00:00:08,400 --> 00:00:11,600
And now we're about to really build
and train

5
00:00:11,600 --> 00:00:14,800
the simple linear regression
model on the training set of course.

6
00:00:14,800 --> 00:00:15,100
Right.

7
00:00:15,100 --> 00:00:19,300
So remember that we splitted that data set
into the training set and the test set.

8
00:00:19,533 --> 00:00:23,566
The training set will be used to train
R simple linear regression model.

9
00:00:23,566 --> 00:00:26,300
And the test set
will be used to evaluate it.

10
00:00:26,300 --> 00:00:28,266
So now we have to start with the training.

11
00:00:28,266 --> 00:00:29,300
And so there we go.

12
00:00:29,300 --> 00:00:31,500
Let's create a new code cell.

13
00:00:31,500 --> 00:00:35,666
And let's implement
the very simple linear regression model.

14
00:00:36,300 --> 00:00:36,600
All right.

15
00:00:36,600 --> 00:00:40,700
So the first thing we'll have to do
is to import the right class

16
00:00:40,866 --> 00:00:43,866
with which we're going to build this
simple linear regression model.

17
00:00:43,900 --> 00:00:49,033
Because indeed we could either implement
it from scratch or we could use libraries.

18
00:00:49,166 --> 00:00:52,166
And of course we're going to use libraries
because I want to provide

19
00:00:52,266 --> 00:00:56,333
a very clear coat template
which allows you to build

20
00:00:56,333 --> 00:00:59,333
any simple linear
regression models in a flashlight.

21
00:00:59,366 --> 00:01:02,533
And this library that we're going to use
is scikit learn,

22
00:01:02,700 --> 00:01:07,466
from which we're going to get access
to a certain module called linear model.

23
00:01:07,466 --> 00:01:11,766
And from this module we're going to call
a certain class called linear regression.

24
00:01:12,000 --> 00:01:13,966
And our simple linear regression model

25
00:01:13,966 --> 00:01:17,733
which we're going to build
will be exactly an instance of this class.

26
00:01:17,733 --> 00:01:20,633
Right. It will be an object of this class.

27
00:01:20,633 --> 00:01:21,733
All right. So let's do this.

28
00:01:21,733 --> 00:01:26,033
Let's start by importing from,
you know, from the scikit

29
00:01:26,033 --> 00:01:29,500
learn library
which has the code name sklearn.

30
00:01:29,766 --> 00:01:34,200
So from sklearn then as we said we're
going to get access to a certain module.

31
00:01:34,200 --> 00:01:39,366
So we have to add a dot here
and which is linear model.

32
00:01:39,600 --> 00:01:42,400
And from this linear module of the scikit

33
00:01:42,400 --> 00:01:46,966
learn library
well we're going to import the linear

34
00:01:47,266 --> 00:01:52,366
regress session class exactly
this one linear regression.

35
00:01:52,766 --> 00:01:57,600
And then as we said this simple linear
regression model which we're going to

36
00:01:57,600 --> 00:02:03,566
build will be an instance or an object
of this linear regression class.

37
00:02:03,800 --> 00:02:07,266
And therefore here
we have to create a new variable

38
00:02:07,500 --> 00:02:11,566
which will be exactly this instance
of the linear regression class.

39
00:02:11,800 --> 00:02:13,066
And we're going to call this object.

40
00:02:13,066 --> 00:02:16,500
We can call it by any name
but we're going to call it regressor

41
00:02:16,800 --> 00:02:19,666
because indeed we are doing right now
regression.

42
00:02:19,666 --> 00:02:20,100
Right.

43
00:02:20,100 --> 00:02:23,833
I remind a big difference
between regression and classification.

44
00:02:24,033 --> 00:02:25,966
Regression is when you have to predict

45
00:02:25,966 --> 00:02:30,100
a continuous real value, like a salary,
as we're about to do.

46
00:02:30,400 --> 00:02:34,766
And classification is when you have to
predict a category or, you know, a class,

47
00:02:34,900 --> 00:02:38,400
which we will do in part
three classification or all right.

48
00:02:38,733 --> 00:02:41,733
So regressor that's a new variable
which at the same time

49
00:02:41,733 --> 00:02:45,333
will become the object of the linear
regression class.

50
00:02:45,333 --> 00:02:49,766
And you can exactly see this object
as the linear regression model itself.

51
00:02:49,766 --> 00:02:52,900
You know, I remind that
a class allows you to implement

52
00:02:53,100 --> 00:02:55,566
a couple of instructions
to build something.

53
00:02:55,566 --> 00:02:56,633
And well, this linear

54
00:02:56,633 --> 00:03:00,666
regression class built exactly
this simple linear regression model.

55
00:03:00,666 --> 00:03:01,033
All right.

56
00:03:01,033 --> 00:03:04,500
So you have to see this regressor object
as exactly this model.

57
00:03:04,933 --> 00:03:05,800
Right. So regressor.

58
00:03:05,800 --> 00:03:08,233
And then to create an object of a class.

59
00:03:08,233 --> 00:03:09,800
Well there's nothing more simple.

60
00:03:09,800 --> 00:03:14,500
You just have to call the class itself
linear regression.

61
00:03:14,800 --> 00:03:17,333
And then add some parentheses
and that's it.

62
00:03:17,333 --> 00:03:20,400
Usually there are some parameters inside
that we can implement.

63
00:03:20,600 --> 00:03:22,900
But here you don't have to enter anything.

64
00:03:22,900 --> 00:03:25,200
This will just create a simple linear
regression model.

65
00:03:25,200 --> 00:03:27,000
And it is so simple that usually

66
00:03:27,000 --> 00:03:29,800
we don't have to play too much
with the parameters.

67
00:03:29,800 --> 00:03:30,533
All right.

68
00:03:30,533 --> 00:03:35,166
And that line of code directly creates
the simple linear regression model.

69
00:03:35,466 --> 00:03:37,400
And that's only the building part.

70
00:03:37,400 --> 00:03:39,633
You know we actually get a model.

71
00:03:39,633 --> 00:03:42,400
But now of course
we have to train it on the training set.

72
00:03:42,400 --> 00:03:46,000
And therefore we have to connect it
in some way to the training set

73
00:03:46,166 --> 00:03:50,100
and the action or,
you know, the function that connects it

74
00:03:50,100 --> 00:03:53,300
in order to train
it is called the fit function.
