1
00:00:00,267 --> 00:00:04,690
[MUSIC]

2
00:00:04,690 --> 00:00:08,326
We've discussed a little bit of
this loan application domain and

3
00:00:08,326 --> 00:00:11,340
showed you a little example
of the decision tree.

4
00:00:11,340 --> 00:00:14,560
Let's dive in and explore the decision
tree a little bit further.

5
00:00:15,600 --> 00:00:21,100
So, I start for my loan application and
I could go down one branch or

6
00:00:21,100 --> 00:00:25,980
one path of this decision tree and
say If this application somebody would

7
00:00:25,980 --> 00:00:30,650
file credit that's fair and
is a short-term loan,

8
00:00:30,650 --> 00:00:34,060
let's say three year loan, then this
is considered a risky application.

9
00:00:35,090 --> 00:00:38,300
If I take another path of
the decision tree, let's say,

10
00:00:38,300 --> 00:00:42,070
I'll look at somebody whose credit
is poor but the income is high but

11
00:00:42,070 --> 00:00:45,220
it's a short term loan again,
I might consider it a risky application.

12
00:00:46,720 --> 00:00:50,750
In general I'm going to
be given an imput xi.

13
00:00:50,750 --> 00:00:54,370
In this case it has assignments to each
one of those input features that say

14
00:00:54,370 --> 00:00:59,380
credit poor, income high, and
term five years longer term loan.

15
00:00:59,380 --> 00:01:04,660
And then I traverse the path down the tree
to make a prediction which in this case

16
00:01:04,660 --> 00:01:06,560
y hat i would be safe.

17
00:01:06,560 --> 00:01:09,860
So this is the model predicts
to be a safe loan to make.

18
00:01:11,790 --> 00:01:17,600
So just in generally, the task that we're
looking at is given some input, xi, I'm

19
00:01:17,600 --> 00:01:22,910
going to traverse the decision tree down
the particular branches of that input,

20
00:01:22,910 --> 00:01:27,170
which in the example that we just did
was a traversal that looks like this.

21
00:01:27,170 --> 00:01:29,750
And then I'm going to make
a prediction y-hat i.

22
00:01:29,750 --> 00:01:36,063
Which might be that this loan is safe and
so our goal is to learn that tree

23
00:01:36,063 --> 00:01:41,104
T(x) that let's us make
predictions about our data.

24
00:01:41,104 --> 00:01:45,529
[MUSIC]