1
00:00:00,000 --> 00:00:04,267
[MUSIC]

2
00:00:04,267 --> 00:00:07,702
Okay so in the first course
of this specialization,

3
00:00:07,702 --> 00:00:11,930
we introduced this block diagram or
flow chart for regression.

4
00:00:11,930 --> 00:00:14,230
As well as a bunch of other
machine learning tasks, but

5
00:00:14,230 --> 00:00:15,770
I just want to walk through this again.

6
00:00:17,290 --> 00:00:19,590
So we're going to assume that
we have some training data.

7
00:00:19,590 --> 00:00:21,710
Okay, so all of a sudden I've
said the word training data,

8
00:00:21,710 --> 00:00:23,800
and I hadn't said that to this point.

9
00:00:23,800 --> 00:00:26,950
That's a topic we discussed
at a very high level

10
00:00:26,950 --> 00:00:28,560
in the first course of the specialization,

11
00:00:28,560 --> 00:00:33,660
and we're gonna discuss the concepts of
training data, test data, validation sets,

12
00:00:33,660 --> 00:00:38,720
lots of other ideas about fitting
our models and accessing our fits.

13
00:00:38,720 --> 00:00:42,000
And chosing between models and
all these different topics.

14
00:00:42,000 --> 00:00:44,080
We're gonna discuss that in this course.

15
00:00:44,080 --> 00:00:48,800
But for right now, if you don't know what
it means to have training data, go back to

16
00:00:48,800 --> 00:00:54,040
the first course of this specialization,
watch that video or look it up online.

17
00:00:54,040 --> 00:00:57,770
For now, I'm gonna assume that you
know that we've some training data and

18
00:00:57,770 --> 00:01:02,610
we're using our training data
to fit a specific model.

19
00:01:02,610 --> 00:01:06,510
And for the rest of this module,
that's gonna suffice,

20
00:01:06,510 --> 00:01:08,360
we're only looking at training data.

21
00:01:08,360 --> 00:01:11,620
And all the other discussion
about validation sets, test sets,

22
00:01:11,620 --> 00:01:13,780
and everything like this,
choosing between models,

23
00:01:13,780 --> 00:01:17,930
assessing the fit of the models,
that's going to come later in this course.

24
00:01:17,930 --> 00:01:21,040
Okay, so when I talk about data
in the rest of this module,

25
00:01:21,040 --> 00:01:23,453
I'm assuming we're talking
about just our training set,

26
00:01:23,453 --> 00:01:27,870
that we've already done some split and
we have our training data, okay.

27
00:01:28,900 --> 00:01:31,740
Think I've said enough on that.

28
00:01:31,740 --> 00:01:37,109
So, we have our training data and
in this case,

29
00:01:37,109 --> 00:01:41,671
that's gonna be some table of house id,

30
00:01:41,671 --> 00:01:47,329
house square feet and
then the house sales price.

31
00:01:51,844 --> 00:01:56,210
And we have this big table
of all these quantities.

32
00:01:56,210 --> 00:02:00,700
And then what we're going to do,
is extract some features.

33
00:02:00,700 --> 00:02:04,510
So maybe, actually, at this point instead
of specifically saying how square feet,

34
00:02:04,510 --> 00:02:07,150
I can just say that there's
some set of house attributes.

35
00:02:07,150 --> 00:02:12,300
We might have things in
addition to square feet.

36
00:02:12,300 --> 00:02:17,790
And so one step we're gonna have to do is
figure out what input we're gonna use for

37
00:02:17,790 --> 00:02:19,570
our regression model.

38
00:02:19,570 --> 00:02:22,340
We're gonna talk about more general
features later in the course, but for

39
00:02:22,340 --> 00:02:27,340
now we're just assuming a simple setup
where we're gonna choose one of our house

40
00:02:27,340 --> 00:02:31,810
attributes, call that the input to this
regression model, and work from there.

41
00:02:31,810 --> 00:02:36,880
So, we've talked about using
square feet as our selected input

42
00:02:38,140 --> 00:02:40,350
and then we're gonna use
our Machine Learning model,

43
00:02:40,350 --> 00:02:46,700
which is regression,
to predict our house sales prices.

44
00:02:46,700 --> 00:02:51,936
So y hat represents our predicted,

45
00:02:53,654 --> 00:02:57,416
house sales price.

46
00:02:57,416 --> 00:02:59,630
And how are we predicting it?

47
00:02:59,630 --> 00:03:05,346
Well, based on some estimated line or
curve,

48
00:03:05,346 --> 00:03:11,369
so this is our,
f hat is our estimated function.

49
00:03:14,701 --> 00:03:21,040
That's fit from our data set, or,
specifically our training data set.

50
00:03:21,040 --> 00:03:25,540
And how are we gonna determine f hat?

51
00:03:25,540 --> 00:03:28,540
Or how are we gonna
estimate this function?

52
00:03:28,540 --> 00:03:33,770
Well, first we're gonna need to
describe some quality metric that says

53
00:03:33,770 --> 00:03:35,690
I should describe what Y is.

54
00:03:35,690 --> 00:03:38,510
I already mentioned this earlier,
but let's write it down explicitly.

55
00:03:38,510 --> 00:03:41,520
This is the sales price, the actual

56
00:03:43,450 --> 00:03:48,290
sales price of our houses, and
we're gonna compare the actual

57
00:03:48,290 --> 00:03:53,720
sales price to the predicted sales
price using the any given F hat.

58
00:03:53,720 --> 00:03:55,290
We're gonna say how well did we do?

59
00:03:56,770 --> 00:03:58,540
And that's what the quality metric is.

60
00:03:58,540 --> 00:04:03,828
So there's gonna be some error
in our predicted values.

61
00:04:09,285 --> 00:04:13,966
And the machine learning algorithm
we're gonna use to fit these

62
00:04:13,966 --> 00:04:18,140
regression models is gonna
try to minimize that error.

63
00:04:18,140 --> 00:04:22,600
So it's gonna search over all these
functions to reduce the error

64
00:04:24,480 --> 00:04:26,550
in these predicted values.

65
00:04:26,550 --> 00:04:29,008
Okay, so
this is our overall flow chart and

66
00:04:29,008 --> 00:04:33,246
we're gonna walk through the various
components of this throughout this

67
00:04:33,246 --> 00:04:36,202
module and
in more depth in the rest of this course.

68
00:04:36,202 --> 00:04:36,702
[MUSIC]