1
00:00:00,139 --> 00:00:04,573
[MUSIC]

2
00:00:04,573 --> 00:00:07,100
Now let's delve into what this
course is gonna be about.

3
00:00:08,530 --> 00:00:09,670
In the Foundations course,

4
00:00:09,670 --> 00:00:14,070
we talked about how machine learning is
about deriving intelligence from data.

5
00:00:14,070 --> 00:00:14,800
And in this course,

6
00:00:14,800 --> 00:00:18,670
the machine learning method that
we're gonna focus on is regression.

7
00:00:18,670 --> 00:00:23,820
And in particular, what regression is
gonna assume is that we have some features

8
00:00:23,820 --> 00:00:28,790
that are derived from our data that
are the input to our regression model.

9
00:00:28,790 --> 00:00:32,490
And then our goal will be to predict
some continuous valued output or

10
00:00:32,490 --> 00:00:34,690
response to the input.

11
00:00:34,690 --> 00:00:40,390
The way we're gonna do this is by learning
a relationship between our inputs x and

12
00:00:40,390 --> 00:00:41,040
this output y.

13
00:00:42,260 --> 00:00:45,900
For example, maybe you're interested
in how taking this machine learning

14
00:00:45,900 --> 00:00:49,600
specialization is going to pay off for
you in the end.

15
00:00:49,600 --> 00:00:53,920
So you're sitting here,
you're doing a lot of really hard work and

16
00:00:53,920 --> 00:00:56,068
you wonder where this
is going to land you.

17
00:00:56,068 --> 00:01:00,510
Well, maybe a question you might be
interested in is what will your salary be

18
00:01:00,510 --> 00:01:03,430
after taking this specialization.

19
00:01:03,430 --> 00:01:06,480
And so we can think about
predicting what your salary is

20
00:01:06,480 --> 00:01:09,690
based on things like what your
performance was in the various courses,

21
00:01:09,690 --> 00:01:12,320
the quality of your capstone project,
how many

22
00:01:12,320 --> 00:01:16,650
forum responses you are participating in,
and different features like this.

23
00:01:16,650 --> 00:01:18,780
So this would be the input
to the regression model and

24
00:01:18,780 --> 00:01:22,320
the prediction, the output that
we're trying to predict would be

25
00:01:22,320 --> 00:01:25,120
your expected salary at the end
of this specialization.

26
00:01:26,140 --> 00:01:28,830
Another example is predicting
the price of a stock.

27
00:01:28,830 --> 00:01:32,740
And to form this prediction
maybe we expect that this would

28
00:01:32,740 --> 00:01:38,070
depend on the past history of the stock,
as well as perhaps recent news events,

29
00:01:38,070 --> 00:01:43,950
in addition to the trends in
other related commodities.

30
00:01:43,950 --> 00:01:45,310
Or maybe you tweeted something, and

31
00:01:45,310 --> 00:01:48,250
you wanna know how many people
are gonna retweet what you tweeted.

32
00:01:48,250 --> 00:01:51,010
Well this might depend on
how many followers you have,

33
00:01:51,010 --> 00:01:53,360
how many followers your followers have,

34
00:01:53,360 --> 00:01:58,620
local structure of your follower network,
what hash tags you used,

35
00:01:58,620 --> 00:02:02,100
how many retweets you've had in the past,
and other features like this.

36
00:02:02,100 --> 00:02:05,320
Another example that we're going
to talk about in this course is

37
00:02:05,320 --> 00:02:07,560
a really cool example
of reading your mind.

38
00:02:07,560 --> 00:02:12,725
Where you go and you get some kind
of brain scan, could be FMRI or

39
00:02:12,725 --> 00:02:17,310
MEG and for our sake we're just going to
think of it as producing an image of your

40
00:02:17,310 --> 00:02:21,200
brain even though the truth is it
produces something more complicated.

41
00:02:21,200 --> 00:02:25,400
But we can think of all
the different pixel intensities

42
00:02:25,400 --> 00:02:30,030
as inputs to a regression model
where the goal of the output is to

43
00:02:30,030 --> 00:02:34,250
predict whether you felt happy or sad
in response to something you were shown

44
00:02:34,250 --> 00:02:36,020
when you were getting that brain scan.

45
00:02:36,020 --> 00:02:39,670
So it's reading your mind because we want
to guess how you're feeling just from

46
00:02:39,670 --> 00:02:42,040
an image of your brain.

47
00:02:42,040 --> 00:02:42,785
But in this course,

48
00:02:42,785 --> 00:02:46,120
we're gonna focus in on a case
study of predicting house prices.

49
00:02:47,430 --> 00:02:53,620
So in particular, a question we're gonna
ask is, what's the value of a given house?

50
00:02:53,620 --> 00:02:55,080
Maybe you wanna sell your house and

51
00:02:55,080 --> 00:02:57,990
you wanna figure out how
much to list that house for.

52
00:02:57,990 --> 00:03:03,130
And so we're gonna derive this
intelligence by looking at some data.

53
00:03:03,130 --> 00:03:06,190
And the data we're gonna look
at include other house sales.

54
00:03:06,190 --> 00:03:09,960
So we're gonna have the sales price
associated with a bunch of other houses,

55
00:03:09,960 --> 00:03:14,040
as well as the house attributes
of these other houses, and

56
00:03:14,040 --> 00:03:17,910
from these inputs, the house attributes,
we're gonna learn this relationship

57
00:03:17,910 --> 00:03:22,520
between house attributes and the output,
which is the sales price, and

58
00:03:22,520 --> 00:03:28,000
use this learned model in order to make
the prediction of the value of your house.

59
00:03:28,000 --> 00:03:31,123
And this course is all about how to form

60
00:03:31,123 --> 00:03:35,521
this relationship between the input and
the output.

61
00:03:35,521 --> 00:03:39,609
[MUSIC]