﻿1
00:00:00,166 --> 00:00:02,800
Hello and welcome to this art tutorial.

2
00:00:02,800 --> 00:00:06,266
So in the previous tutorial
we predicted our test results

3
00:00:06,266 --> 00:00:10,500
and now in this tutorial
we are going to evaluate those predictions

4
00:00:10,866 --> 00:00:14,133
by making the confusion matrix,
which will count

5
00:00:14,133 --> 00:00:17,700
the number of correct predictions
and the number of incorrect predictions.

6
00:00:18,200 --> 00:00:18,866
So let's do it.

7
00:00:18,866 --> 00:00:20,900
Let's make the matrix. It's very simple.

8
00:00:20,900 --> 00:00:23,166
It will just take us one line.

9
00:00:23,166 --> 00:00:26,166
So let's call it CM equals.

10
00:00:26,633 --> 00:00:30,866
Then very practically we're going
to use the table function in R.

11
00:00:31,533 --> 00:00:35,433
And in this table
I will first input the real values

12
00:00:35,433 --> 00:00:39,466
which are the test set brackets.

13
00:00:39,466 --> 00:00:42,933
And then I have to select
the columns of the real results which is

14
00:00:43,466 --> 00:00:47,700
if we go to the test set this column
because this column contains

15
00:00:47,700 --> 00:00:48,600
the real results.

16
00:00:48,600 --> 00:00:53,533
Whether the user bought yes or no,
the SUV and this column has index three.

17
00:00:53,733 --> 00:00:55,566
So here I will put

18
00:00:56,566 --> 00:00:59,266
comma and three.

19
00:00:59,266 --> 00:01:00,833
All right. So that's my first argument.

20
00:01:00,833 --> 00:01:02,300
That's the real values.

21
00:01:02,300 --> 00:01:06,300
And then as the second argument
I'm going to input the predicted values

22
00:01:06,633 --> 00:01:09,633
which is of course the y vector here.

23
00:01:09,833 --> 00:01:13,066
So here I'm going to add y prep.

24
00:01:13,766 --> 00:01:17,000
And I don't have to select any index
because Whitbread is already

25
00:01:17,000 --> 00:01:19,033
the vector of predictions.

26
00:01:19,033 --> 00:01:24,266
So in short this is the vector
of real values for all the observations.

27
00:01:24,633 --> 00:01:27,533
And this is the vector of predictions

28
00:01:27,533 --> 00:01:30,533
for the same observations
as in this vector

29
00:01:31,433 --> 00:01:33,033
okay. And that's it.

30
00:01:33,033 --> 00:01:34,966
The confusion matrix is ready.

31
00:01:34,966 --> 00:01:37,966
So now let's select this line and execute.

32
00:01:38,500 --> 00:01:41,500
Here it is table created.

33
00:01:41,600 --> 00:01:44,533
Now let's go to the console
and have a look at it

34
00:01:44,533 --> 00:01:47,266
CM enter.

35
00:01:47,266 --> 00:01:47,933
And here it is.

36
00:01:48,933 --> 00:01:50,700
The most
important thing to understand here

37
00:01:50,700 --> 00:01:55,700
is that the 57 and the 26
here are the correct predictions.

38
00:01:56,166 --> 00:01:59,866
And the ten and the seven here
are the incorrect predictions.

39
00:02:00,733 --> 00:02:05,333
So what's interesting here at first sight
is that the classifier made

40
00:02:05,666 --> 00:02:09,000
57 plus 26 equals 83.

41
00:02:09,300 --> 00:02:14,233
Correct predictions, and ten plus
seven equals 17 incorrect predictions.

42
00:02:14,766 --> 00:02:15,166
All right.

43
00:02:15,166 --> 00:02:17,900
So 17 incorrect
predictions on the test set.

44
00:02:17,900 --> 00:02:20,133
That's not bad. But we can do better.

45
00:02:20,133 --> 00:02:23,133
And we will do better
with other classifiers.

46
00:02:23,400 --> 00:02:25,800
You will see that in the next sections
okay.

47
00:02:25,800 --> 00:02:28,333
So we're done with the confusion matrix.

48
00:02:28,333 --> 00:02:30,866
And finally
now it's time for the best part

49
00:02:30,866 --> 00:02:34,666
because in the next tutorial we will be
graphically looking at our results

50
00:02:34,933 --> 00:02:39,500
because we will plot this very cool chart
that will allow us to make an awesome

51
00:02:39,500 --> 00:02:41,300
interpretation of the results.

52
00:02:41,300 --> 00:02:43,700
So I look forward to seeing you
in the next tutorial

53
00:02:43,700 --> 00:02:46,133
where we will make this chart.

54
00:02:46,133 --> 00:02:47,833
Until then, enjoy machine learning.
