1
00:00:00,000 --> 00:00:04,298
[MUSIC]

2
00:00:04,298 --> 00:00:07,320
And so let's actually use the model.

3
00:00:08,430 --> 00:00:15,960
So, next we've now trained a liner
regression model, let's evaluate it.

4
00:00:15,960 --> 00:00:20,834
So, what we're gonna do next is

5
00:00:20,834 --> 00:00:25,540
#Evaluate the simple model.

6
00:00:25,540 --> 00:00:26,680
So how to evaluate it?

7
00:00:26,680 --> 00:00:28,900
We're gonna look at the test data.

8
00:00:28,900 --> 00:00:32,411
So, remember we had the split test data.

9
00:00:32,411 --> 00:00:34,980
Let's understand test data a little bit.

10
00:00:34,980 --> 00:00:37,380
So, for example, let's print.

11
00:00:38,940 --> 00:00:43,090
For the test data, for

12
00:00:43,090 --> 00:00:47,210
the price column,
what's the average price?

13
00:00:47,210 --> 00:00:48,330
What's the mean price?

14
00:00:50,050 --> 00:00:53,230
So this just computes the average
price and the average price for

15
00:00:53,230 --> 00:00:56,790
the test data, for
this data from Seattle is $543,000.

16
00:00:56,790 --> 00:01:03,230
That's how average house costs,
it's pretty expensive actually.

17
00:01:03,230 --> 00:01:06,760
Now, we've built a square foot model,
and so

18
00:01:06,760 --> 00:01:11,080
what we want to do is evaluate
it on this test data.

19
00:01:11,080 --> 00:01:16,080
So we're going to take the sqft_model

20
00:01:16,080 --> 00:01:21,150
that we built and we're gonna call
what's called evaluate function.

21
00:01:21,150 --> 00:01:25,910
Which can take a test data set and
print out or

22
00:01:25,910 --> 00:01:30,400
return some statistics of
how well that fit is doing.

23
00:01:32,370 --> 00:01:34,310
So let's do that.

24
00:01:36,140 --> 00:01:40,240
So, I'm actually going to type,
print in the beginning,

25
00:01:40,240 --> 00:01:42,860
because it formats it a little bit nicer.

26
00:01:42,860 --> 00:01:50,850
And you'll see that the maximal error
over all test houses was 4.1 million.

27
00:01:50,850 --> 00:01:55,040
So there was one house that was
an outlier, it was really badly predicted.

28
00:01:55,040 --> 00:01:59,856
And average error, so the RMSE,
so the root means squared error,

29
00:01:59,856 --> 00:02:01,496
we talked about this.

30
00:02:01,496 --> 00:02:06,941
Emily talked about this with
us during the lectures,

31
00:02:06,941 --> 00:02:11,550
is $255,000, so that's the RMSE.

32
00:02:11,550 --> 00:02:18,103
So we built this simple model,
we tested, has pretty high RMSE,

33
00:02:18,103 --> 00:02:24,791
but let's look at some predictions
it tries to make from the data.

34
00:02:24,791 --> 00:02:29,099
[MUSIC]