1
00:00:00,300 --> 00:00:01,133
Are you ready?

2
00:00:01,133 --> 00:00:02,033
Let's do this.

3
00:00:02,033 --> 00:00:06,700
Let's start implementing
our multiple linear regression model.

4
00:00:07,000 --> 00:00:08,366
So I just double clicked on it.

5
00:00:08,366 --> 00:00:11,733
If you like Google Collaboratory,
feel free to open this link with me.

6
00:00:11,733 --> 00:00:13,366
Open with Google Collaboratory.

7
00:00:13,366 --> 00:00:15,700
And if you don't like Google
Collaboratory, that's totally fine.

8
00:00:15,700 --> 00:00:18,266
You can open this file
with Jupyter Notebook,

9
00:00:18,266 --> 00:00:21,266
but from your folder,
which you downloaded on your machine.

10
00:00:21,333 --> 00:00:22,866
All right, choose your favorite.

11
00:00:22,866 --> 00:00:24,266
And here we go.

12
00:00:24,266 --> 00:00:27,600
Let's implement
our multiple linear regression model.

13
00:00:28,800 --> 00:00:29,333
All right.

14
00:00:29,333 --> 00:00:33,066
So first remember that this file is in

15
00:00:33,066 --> 00:00:36,466
read only mode
which means that we can't modify it.

16
00:00:36,566 --> 00:00:41,033
But no worries we're going to create
a copy right away by going to file here.

17
00:00:41,033 --> 00:00:44,033
And then click save a copy in drive.

18
00:00:44,166 --> 00:00:48,833
This will as you can see, create a copy
in which we will be able to re-implement

19
00:00:49,000 --> 00:00:52,100
this multiple linear
regression model from scratch.

20
00:00:52,266 --> 00:00:56,600
Because indeed, I remind that
this course is an action based course

21
00:00:56,833 --> 00:00:59,733
in which I want you to take action
as much as you can.

22
00:00:59,733 --> 00:01:01,433
And therefore,
we are going to re-implement

23
00:01:01,433 --> 00:01:04,400
this whole model from scratch
together, step by step.

24
00:01:04,400 --> 00:01:07,933
And I really want you to code with me
at the same time so that, you know,

25
00:01:08,033 --> 00:01:11,466
the practical skill can really be
well integrated in your head.

26
00:01:11,666 --> 00:01:13,200
Okay, so let's do this.

27
00:01:13,200 --> 00:01:18,466
Let's first start by removing old code
cells here and only the code cells,

28
00:01:18,466 --> 00:01:23,100
not the text cells because I want to keep
that well highlighted structure

29
00:01:23,866 --> 00:01:26,166
for this implementation.

30
00:01:26,166 --> 00:01:29,066
So here I'm just removing all the code
cells.

31
00:01:29,066 --> 00:01:30,900
There we go. And perfect.

32
00:01:30,900 --> 00:01:33,100
So this is the whole structure
of this implementation.

33
00:01:33,100 --> 00:01:34,666
We can have a look at it here.

34
00:01:34,666 --> 00:01:38,800
And indeed
that's why you know I wanted to brainstorm

35
00:01:38,800 --> 00:01:43,033
on the dataset with you first
before showing you the structure

36
00:01:43,166 --> 00:01:46,866
in order to make you think what we must do
in the data preprocessing phase.

37
00:01:46,866 --> 00:01:47,766
And as we said,

38
00:01:47,766 --> 00:01:52,100
we need first to import the libraries,
then import the data set, that's for sure.

39
00:01:52,100 --> 00:01:53,133
And then here we go.

40
00:01:53,133 --> 00:01:55,966
We need to encode the categorical data.

41
00:01:55,966 --> 00:02:00,833
And more specifically that state column
which contains needs three categories.

42
00:02:01,366 --> 00:02:04,366
And then of course we split the data
set into the training set and test it.

43
00:02:04,566 --> 00:02:05,633
That's a must.

44
00:02:05,633 --> 00:02:08,366
And that will close the data
preprocessing phase.

45
00:02:08,366 --> 00:02:12,266
And we will all be ready to start training
while first building

46
00:02:12,266 --> 00:02:16,200
and then training the multiple linear
regression model on the training set.

47
00:02:16,500 --> 00:02:19,500
And by doing this,
our model will understand the correlations

48
00:02:19,500 --> 00:02:24,500
between all these spans of, you know,
those stories and their generated profit.

49
00:02:24,700 --> 00:02:25,633
And so there we go.

50
00:02:25,633 --> 00:02:31,200
We will get a smart model which we will
be able to use on new observations.

51
00:02:31,366 --> 00:02:35,133
And that's exactly what we'll do
as a last step here to predict the test

52
00:02:35,133 --> 00:02:35,566
set results.