1
00:00:00,008 --> 00:00:04,224
[MUSIC]

2
00:00:04,224 --> 00:00:07,260
So we keep talking about
this housing application.

3
00:00:07,260 --> 00:00:12,140
And it's really a nice intuitive way
to describe the different methods in

4
00:00:12,140 --> 00:00:13,960
regression that we're gonna
be talking about, and

5
00:00:13,960 --> 00:00:15,700
that we have been talking about.

6
00:00:15,700 --> 00:00:20,480
But regression, like we've mentioned, is
much, much, much more widely applicable.

7
00:00:20,480 --> 00:00:24,040
And this notation of
capturing seasonality also

8
00:00:24,040 --> 00:00:27,660
appears in lots of applications
beyond just housing.

9
00:00:27,660 --> 00:00:31,390
And so I wanted to spend a little time
talking about other places where we see

10
00:00:31,390 --> 00:00:36,370
this seasonality or
seasonal effects coming into play.

11
00:00:36,370 --> 00:00:39,870
And one, which makes a lot of sense,
is if you're doing weather modeling.

12
00:00:39,870 --> 00:00:43,350
Let's say you're trying to predict
the temperature or rainfall.

13
00:00:43,350 --> 00:00:44,770
Well, if you're thinking
about temperature,

14
00:00:44,770 --> 00:00:47,805
well there's variation in
temperature across a day.

15
00:00:47,805 --> 00:00:53,230
It's hotter during the daytime hours and
cooler during the night.

16
00:00:53,230 --> 00:00:57,670
But of course, it's also hotter in
the summer and cooler in the winter.

17
00:00:57,670 --> 00:01:02,360
So there's actually this seasonality
at different time scales.

18
00:01:02,360 --> 00:01:05,450
So in addition to just
having that one sine and

19
00:01:05,450 --> 00:01:09,600
cosine that we showed in that
housing model, where in that case,

20
00:01:09,600 --> 00:01:14,050
we're just just looking at this
monthly effect repeating every year.

21
00:01:14,050 --> 00:01:17,105
Well, for weather modelling,
if you're predicting temperature,

22
00:01:17,105 --> 00:01:19,245
you might wanna add in a sine and

23
00:01:19,245 --> 00:01:22,765
cosine functions at different frequencies
to capture the fact that there are these

24
00:01:22,765 --> 00:01:28,105
daily effects as well as these monthly
effects, and maybe other effects as well.

25
00:01:29,575 --> 00:01:33,725
Also you might be thinking about Flu
monitoring, so you wanna think about

26
00:01:33,725 --> 00:01:38,590
monitoring the incidence rate of flu and,
for example here, the picture I'm

27
00:01:38,590 --> 00:01:42,170
showing is in the United States in
a whole bunch of different regions.

28
00:01:42,170 --> 00:01:47,450
And if we look at any one of those regions
and we look at rate of flu over time,

29
00:01:47,450 --> 00:01:51,615
well of course there are gonna
be peaks during flu season and

30
00:01:51,615 --> 00:01:53,795
valleys during the off months.

31
00:01:53,795 --> 00:01:58,585
And so you see this kind of seasonal
pattern in flu monitoring and

32
00:01:58,585 --> 00:02:03,065
lots of other types of
health monitoring like this.

33
00:02:04,680 --> 00:02:08,860
And you'll also see it in the things
like E-commerce, for example, Amazon

34
00:02:08,860 --> 00:02:14,450
is really interested in being able to
stock their inventory pretty accurately.

35
00:02:14,450 --> 00:02:18,100
And if you're thinking about selling
jackets, you have some warehouse here in

36
00:02:18,100 --> 00:02:24,880
the US and you wanna figure out how many
snow jackets or ski jackets to stock.

37
00:02:24,880 --> 00:02:29,600
Well, of course, you're gonna wanna stock
more in the winter months than you would

38
00:02:29,600 --> 00:02:33,680
in the summer months because more
people are likely to purchase jackets

39
00:02:33,680 --> 00:02:36,130
in the winter than in the summer.

40
00:02:36,130 --> 00:02:39,550
So that's another place where
seasonality is really important.

41
00:02:39,550 --> 00:02:41,610
And it appears in so many applications.

42
00:02:41,610 --> 00:02:44,470
Another one that you might not
think of is Motion capture,

43
00:02:44,470 --> 00:02:48,450
just trying to model how
a person walks over time.

44
00:02:48,450 --> 00:02:51,780
And if you look at the data, if you
put sensors over a person's body and

45
00:02:51,780 --> 00:02:56,420
look at how they walk, if you take
those recordings, you're gonna get

46
00:02:56,420 --> 00:03:00,500
these kind of up and down, up and
down swings, as the person's going through

47
00:03:00,500 --> 00:03:04,640
their different motions, raising
their knees or walking or their arms.

48
00:03:04,640 --> 00:03:09,175
And so in this plot here, I'm looking at
some center trajectories from a person

49
00:03:09,175 --> 00:03:13,915
wearing a motion capture suit as they're
going through different behaviors, and

50
00:03:13,915 --> 00:03:17,051
you clearly see this type of
seasonality here as well.

51
00:03:17,051 --> 00:03:21,179
[MUSIC]