WEBVTT

00:00:00.090 --> 00:00:03.000
By this point in this class, you
must have realized that a

00:00:03.000 --> 00:00:06.730
lot of revolves around the
concept of conditioning.

00:00:06.730 --> 00:00:09.900
Conditional expectations
play a central role.

00:00:09.900 --> 00:00:13.100
For this reason, it is useful
to revisit this concept and

00:00:13.100 --> 00:00:16.040
view it in a more
abstract manner.

00:00:16.040 --> 00:00:18.780
The basic idea is that the
value of a conditional

00:00:18.780 --> 00:00:23.390
expectation is affected by a
random quantity by the value

00:00:23.390 --> 00:00:27.400
of the random variable Y on
which we are conditioning.

00:00:27.400 --> 00:00:32.750
It is a function of Y and,
therefore, a random variable.

00:00:32.750 --> 00:00:36.130
Based on this observation, we
will redefine the conditional

00:00:36.130 --> 00:00:39.830
expectation as a random variable
and then try to

00:00:39.830 --> 00:00:41.950
understand its properties.

00:00:41.950 --> 00:00:45.430
In particular, we will develop
a formula for the expected

00:00:45.430 --> 00:00:48.950
value of the conditional
expectation.

00:00:48.950 --> 00:00:52.900
This will be what as known
as the law of iterated

00:00:52.900 --> 00:00:55.330
expectations.

00:00:55.330 --> 00:00:58.640
After doing all this, we will
follow a similar program for

00:00:58.640 --> 00:01:00.800
the conditional variance.

00:01:00.800 --> 00:01:02.930
Once more, we will see
that it can be

00:01:02.930 --> 00:01:05.060
viewed as a random variable.

00:01:05.060 --> 00:01:08.400
And then we will relate its
expected value with the

00:01:08.400 --> 00:01:10.590
unconditional variance.

00:01:10.590 --> 00:01:15.940
This will be the so-called
law of total variance.

00:01:15.940 --> 00:01:18.800
As an illustration of the tools
we are introducing in

00:01:18.800 --> 00:01:22.330
this lecture, we will consider
various examples that will

00:01:22.330 --> 00:01:25.880
hopefully clarify the
concepts involved.

00:01:25.880 --> 00:01:29.789
Our final and most important
example will involve the sum

00:01:29.789 --> 00:01:34.490
of a random number of
independent random variables.

00:01:34.490 --> 00:01:37.050
The setting here is more
challenging than the case

00:01:37.050 --> 00:01:40.450
where we add a fixed number
of random variables.

00:01:40.450 --> 00:01:44.050
But by using conditioning,
we will be able to derive

00:01:44.050 --> 00:01:46.360
formulas for the mean
and the variance.