1
00:00:00,006 --> 00:00:04,768
[MUSIC]

2
00:00:04,768 --> 00:00:08,320
So we've now built
a simple scatter plot and

3
00:00:08,320 --> 00:00:12,120
we get an idea of what
our data looks like.

4
00:00:12,120 --> 00:00:13,150
And the question is,

5
00:00:13,150 --> 00:00:19,240
can we use the scatter plot to predict
sales price from square feet of living?

6
00:00:19,240 --> 00:00:20,770
So we're gonna do a simple
regression model.

7
00:00:22,040 --> 00:00:27,558
So let's do that, and we hit SM, and

8
00:00:27,558 --> 00:00:32,721
we're gonna say #Create a simple

9
00:00:32,721 --> 00:00:39,500
regression model of sqft_living to price.

10
00:00:42,490 --> 00:00:44,720
All right, so let's do just that.

11
00:00:46,450 --> 00:00:50,190
So remember from the lectures that
the first thing that you do before you do

12
00:00:50,190 --> 00:00:54,460
anything to your data is to split it
into a training set and a test set,

13
00:00:54,460 --> 00:00:57,720
because you never want to do,
trying training or

14
00:00:57,720 --> 00:01:01,120
learning on the test data, you want
to do that just on the training data.

15
00:01:01,120 --> 00:01:06,696
So let's do that, split, so
I'm gonna take my data and

16
00:01:06,696 --> 00:01:11,918
split it into train_data and
test_data by calling

17
00:01:11,918 --> 00:01:18,088
a function that's called,
that you can apply to an so

18
00:01:18,088 --> 00:01:22,285
it's called the random split function.

19
00:01:22,285 --> 00:01:24,454
Sales.random_split.

20
00:01:25,800 --> 00:01:29,660
By the way, what I did there
was just use tab complete.

21
00:01:29,660 --> 00:01:32,420
So let me just show you that little trick,
just for a second.

22
00:01:32,420 --> 00:01:35,770
So if I just do sales.r and I press Tab,

23
00:01:35,770 --> 00:01:37,108
you'll see there's a few
things I could do.

24
00:01:37,108 --> 00:01:42,450
random_split, read_csv, remove_columns,
rename_columns and so on.

25
00:01:42,450 --> 00:01:44,530
So I'm gonna just do random_split.

26
00:01:44,530 --> 00:01:49,250
And what I'm gonna tell
it is to do a .8 split.

27
00:01:49,250 --> 00:01:55,630
So .8 of the data, 80% is gonna be for
training and 20% is gonna be for testing.

28
00:01:56,930 --> 00:02:01,840
Now, one last thing that will be useful
for your homeworks and in general,

29
00:02:01,840 --> 00:02:07,000
to make sure that you have always the same
results is to set a seed for the split.

30
00:02:07,000 --> 00:02:10,900
Because a random split,
it's a pseudo random number generator.

31
00:02:10,900 --> 00:02:15,290
We can set a seed to it, and we just gonna
set the seed to be any number we want.

32
00:02:15,290 --> 00:02:18,820
For example, it could be 2015,

33
00:02:18,820 --> 00:02:23,500
it could be many things, so
I'm just gonna set it to 0.

34
00:02:23,500 --> 00:02:28,648
So now every time I do this random split,
it splits the data in the same way.

35
00:02:30,366 --> 00:02:34,269
[MUSIC]