1
00:00:00,000 --> 00:00:04,643
[MUSIC]

2
00:00:04,643 --> 00:00:08,071
Now that we've taken a deep
dive into a linear classifier,

3
00:00:08,071 --> 00:00:12,749
namely logistic regression, let's jump
into a different kind of classifier.

4
00:00:12,749 --> 00:00:16,032
One called a decision tree,
which is extremely useful in practice,

5
00:00:16,032 --> 00:00:19,033
especially when it combines
with something called boosting,

6
00:00:19,033 --> 00:00:21,430
which you're going to see
in an upcoming module.

7
00:00:22,490 --> 00:00:25,700
In this module, we're going to jump in
to another really interesting real world

8
00:00:25,700 --> 00:00:29,380
example, which falls more into
the financial data modelling part.

9
00:00:29,380 --> 00:00:32,676
It turns out that decision
trees are extremely useful and

10
00:00:32,676 --> 00:00:37,208
used a lot in the finance industry, so
we're going to jump into something in that

11
00:00:37,208 --> 00:00:41,122
domain to give you a little bit more
experience to that kind of data,

12
00:00:41,122 --> 00:00:44,783
in particular when you look at
evaluating loan applications.

13
00:00:44,783 --> 00:00:46,176
So lets say, I want to buy a house.

14
00:00:46,176 --> 00:00:48,453
I'm excited,
I want to buy this particular house.

15
00:00:48,453 --> 00:00:53,410
To buy it, I don't have all the money,
I need to take a loan from the bank.

16
00:00:53,410 --> 00:00:57,451
So the bank is going to look at some
properties of my history, like my credit,

17
00:00:57,451 --> 00:00:59,206
what has it been like in the past?

18
00:00:59,206 --> 00:01:00,980
How much money I make?

19
00:01:00,980 --> 00:01:06,360
How long my loan is of which time I'm
willing to spend before I pay it off?

20
00:01:06,360 --> 00:01:10,561
And other personal information about me,
like my gender, age and so on.

21
00:01:10,561 --> 00:01:12,720
And he's going to take
that information and

22
00:01:12,720 --> 00:01:17,650
try to make a prediction as to whether
loaning me money is a risky thing or not.

23
00:01:17,650 --> 00:01:20,500
So, let me give you a little bit
more details on the kinds of things

24
00:01:20,500 --> 00:01:24,490
that people measure when they try
to make loan application decisions.

25
00:01:24,490 --> 00:01:26,970
So typically, you look at credit history,

26
00:01:26,970 --> 00:01:30,670
which looks at all the other loans I've
taken in the past and my credit cards and

27
00:01:30,670 --> 00:01:33,330
all that and
have I paid those off on time.

28
00:01:34,980 --> 00:01:37,360
Then we look at my income,
how much money do I make today?

29
00:01:38,470 --> 00:01:42,530
And we'll also look at what's called the
term of the loan, which may be 3 years,

30
00:01:42,530 --> 00:01:43,640
5 years, 15 years,

31
00:01:43,640 --> 00:01:48,220
30 years, which is the period of time
I'm going to take to pay back that loan.

32
00:01:48,220 --> 00:01:51,257
So, let's take a three-year loan
means I pay back within three years.

33
00:01:51,257 --> 00:01:55,625
And finally, a look at information about
the particular individual like my age,

34
00:01:55,625 --> 00:01:57,470
whether married and so on.

35
00:01:57,470 --> 00:01:59,562
So, given this information,

36
00:01:59,562 --> 00:02:04,958
I want to make a prediction as to whether
loaning me money is a risky thing or not.

37
00:02:04,958 --> 00:02:08,720
So, a loan application system
might look something like this.

38
00:02:08,720 --> 00:02:12,703
You get as input all the loan information
that fill up a bunch of forms.

39
00:02:12,703 --> 00:02:15,615
It goes through a system if we
have machine learning system or

40
00:02:15,615 --> 00:02:16,926
maybe minor created one.

41
00:02:16,926 --> 00:02:18,284
Hopefully, machine loan system,

42
00:02:18,284 --> 00:02:20,560
which is going to make a prediction
whether this loan is safe.

43
00:02:20,560 --> 00:02:23,640
It's okay to make that loan or
whether it's risky.

44
00:02:23,640 --> 00:02:27,206
So, some applications is going to be
classified as safe while others will come

45
00:02:27,206 --> 00:02:28,623
in and be classified as risky.

46
00:02:28,623 --> 00:02:33,838
And the bank may be unevenly making
loans to risky applications, but often

47
00:02:33,838 --> 00:02:39,800
to save funds and we can view this loan
application as a classification problem.

48
00:02:39,800 --> 00:02:43,806
I'm given this input information
from my form and let's called the x,

49
00:02:43,806 --> 00:02:45,957
all information about me and my loan.

50
00:02:45,957 --> 00:02:49,090
I pushed it through the classifier
modelm which going to make a decision.

51
00:02:49,090 --> 00:02:53,400
Let say y plus 1, y height equal plus 1,
if it's a safe loan and

52
00:02:53,400 --> 00:02:56,690
y height equals minus 1
if it's a risky loan.

53
00:02:56,690 --> 00:02:59,834
In this module, we're going to use what
is called decision tress classifier.

54
00:02:59,834 --> 00:03:03,908
It might look a little bit like this,
you start the application and

55
00:03:03,908 --> 00:03:07,497
look at some particular feature
of the particular input.

56
00:03:07,497 --> 00:03:09,135
Let's say, what has my credit been like?

57
00:03:09,135 --> 00:03:11,068
If my credit has been excellent,

58
00:03:11,068 --> 00:03:15,220
I just make the loan without looking
at any other information about me.

59
00:03:15,220 --> 00:03:18,889
But if my credit has been only fair,
I look at the term of the loan.

60
00:03:18,889 --> 00:03:22,186
If the term is short,
maybe it's too risky.

61
00:03:22,186 --> 00:03:25,630
But if the loan is long,
maybe I'll take enough time to pay it off.

62
00:03:25,630 --> 00:03:28,220
Now if my credit is poor,
I don't stop there.

63
00:03:28,220 --> 00:03:32,407
I look further where my income is even
if my credit didn't pass being poor by

64
00:03:32,407 --> 00:03:35,995
making a bunch of money,
maybe I'm willing to make a long-term

65
00:03:35,995 --> 00:03:40,000
loan to somebody with poor credit
that makes a bunch of money.

66
00:03:40,000 --> 00:03:43,574
But if you don't make that much money and
your credit is poor,

67
00:03:43,574 --> 00:03:45,094
then you're out of luck.

68
00:03:45,094 --> 00:03:46,729
No loans for you.

69
00:03:46,729 --> 00:03:50,949
[MUSIC]