WEBVTT

00:00:00.360 --> 00:00:03.020
Let us now put to use our
understanding of the

00:00:03.020 --> 00:00:06.000
coin-tossing model and the
associated binomial

00:00:06.000 --> 00:00:07.550
probabilities.

00:00:07.550 --> 00:00:10.210
We will solve the following
problem.

00:00:10.210 --> 00:00:13.910
We have a coin, which
is tossed 10 times.

00:00:13.910 --> 00:00:17.710
And we're told that exactly
three out of the 10 tosses

00:00:17.710 --> 00:00:19.800
resulted in heads.

00:00:19.800 --> 00:00:22.530
Given this information, we would
like to calculate the

00:00:22.530 --> 00:00:27.310
probability that the first
two tosses were heads.

00:00:27.310 --> 00:00:29.940
This is a question of
calculating a conditional

00:00:29.940 --> 00:00:33.670
probability of one event
given another.

00:00:33.670 --> 00:00:37.720
The conditional probability of
event A, namely that the first

00:00:37.720 --> 00:00:41.480
two tosses were heads, given
that another event B has

00:00:41.480 --> 00:00:45.250
occurred, namely that we
had exactly three heads

00:00:45.250 --> 00:00:47.600
out of the 10 tosses.

00:00:47.600 --> 00:00:51.560
However, before we can start
working towards the solution

00:00:51.560 --> 00:00:54.990
to this problem, we need to
specify a probability model

00:00:54.990 --> 00:00:57.080
that we will be working with.

00:00:57.080 --> 00:01:00.070
We need to be explicit about
our assumptions.

00:01:00.070 --> 00:01:03.290
To this effect, let us introduce
the following

00:01:03.290 --> 00:01:04.780
assumptions.

00:01:04.780 --> 00:01:08.140
We will assume that the
different coin tosses are

00:01:08.140 --> 00:01:09.390
independent.

00:01:09.390 --> 00:01:13.270
In addition, we will assume
that each coin toss has a

00:01:13.270 --> 00:01:18.300
fixed probability, p, the same
for each toss, that the

00:01:18.300 --> 00:01:21.560
particular toss results
in heads.

00:01:21.560 --> 00:01:24.090
These are the exact same
assumptions that we made

00:01:24.090 --> 00:01:27.580
earlier when we derived the
binomial probabilities.

00:01:27.580 --> 00:01:31.450
And in particular, we have the
following formula that if we

00:01:31.450 --> 00:01:35.650
have n tosses, the probability
that we obtain exactly k heads

00:01:35.650 --> 00:01:37.880
is given by this expression.

00:01:37.880 --> 00:01:42.550
So now, we have a model in place
and also the tools that

00:01:42.550 --> 00:01:46.400
we can use to analyze this
particular model.

00:01:46.400 --> 00:01:48.289
Let us start working
towards a solution.

00:01:48.289 --> 00:01:51.600
Actually, we will develop two
different solutions and

00:01:51.600 --> 00:01:54.200
compare them at the end.

00:01:54.200 --> 00:01:56.060
The first approach,
which is the

00:01:56.060 --> 00:01:58.710
safest one, is the following.

00:01:58.710 --> 00:02:02.125
Since we want to calculate a
conditional probability, let

00:02:02.125 --> 00:02:05.000
us just start with the
definition of conditional

00:02:05.000 --> 00:02:06.520
probabilities.

00:02:06.520 --> 00:02:10.228
The conditional probability of
an event given another event

00:02:10.228 --> 00:02:13.970
is the probability that both
events happen, divided by the

00:02:13.970 --> 00:02:16.540
probability of the conditioning
event.

00:02:16.540 --> 00:02:21.070
Now, let us specialize to the
particular example that we're

00:02:21.070 --> 00:02:23.740
trying to solve.

00:02:23.740 --> 00:02:27.110
So in the numerator, we're
talking about the probability

00:02:27.110 --> 00:02:31.329
that event A happens and
event B happens.

00:02:31.329 --> 00:02:33.060
What does that mean?

00:02:33.060 --> 00:02:35.440
This means that event
A happens--

00:02:35.440 --> 00:02:39.970
that is, the first two tosses
resulted in heads, which I'm

00:02:39.970 --> 00:02:43.440
going to denote symbolically
this way.

00:02:43.440 --> 00:02:47.610
But in addition to that,
event B happens.

00:02:47.610 --> 00:02:51.260
And event B requires that there
is a total of three

00:02:51.260 --> 00:02:57.720
heads, which means that we
had one more head in

00:02:57.720 --> 00:02:59.250
the remaining tosses.

00:02:59.250 --> 00:03:07.510
So we have one head in tosses
3 all the way to 10.

00:03:15.810 --> 00:03:19.530
As for the denominator, let's
keep it the way it is for now.

00:03:22.410 --> 00:03:26.270
So let's continue with
the numerator.

00:03:26.270 --> 00:03:29.380
We're talking about the
probability of two events

00:03:29.380 --> 00:03:33.970
happening, that the first two
tosses were heads and that in

00:03:33.970 --> 00:03:38.700
tosses 3 up to 10, we had
exactly one head.

00:03:38.700 --> 00:03:42.550
Here comes the independence
assumption.

00:03:42.550 --> 00:03:45.750
Because the different tosses
are independent, whatever

00:03:45.750 --> 00:03:49.650
happens in the first two tosses
is independent from

00:03:49.650 --> 00:03:53.480
whatever happened in
tosses 3 up to 10.

00:03:53.480 --> 00:03:57.270
So the probability of these two
events happening is the

00:03:57.270 --> 00:04:00.420
product of their individual
probabilities.

00:04:00.420 --> 00:04:04.880
So we first have the probability
that the first two

00:04:04.880 --> 00:04:10.170
tosses were heads, which
is p squared.

00:04:10.170 --> 00:04:13.610
And we need to multiply it
with the probability that

00:04:13.610 --> 00:04:17.230
there was exactly one head
in the tosses numbered

00:04:17.230 --> 00:04:19.019
from 3 up to 10.

00:04:19.019 --> 00:04:21.450
These are eight tosses.

00:04:21.450 --> 00:04:25.980
The probability of one head in
eight tosses is given by the

00:04:25.980 --> 00:04:31.820
binomial formula, with k equal
to 1 and n equal to 8.

00:04:31.820 --> 00:04:38.190
So this expression, this part,
becomes 8 choose 1, p to the

00:04:38.190 --> 00:04:44.790
first power times 1 minus
p to the seventh power.

00:04:44.790 --> 00:04:47.560
So this is the numerator.

00:04:47.560 --> 00:04:51.420
The denominator is
easier to find.

00:04:51.420 --> 00:04:53.230
This is the probability
that we had

00:04:53.230 --> 00:04:55.909
three heads in 10 tosses.

00:04:55.909 --> 00:04:57.690
So we just use this formula.

00:04:57.690 --> 00:05:04.220
The probability of three heads
is given by: 10 tosses choose

00:05:04.220 --> 00:05:10.685
three, p to the third, times 1
minus p to the seventh power.

00:05:13.740 --> 00:05:16.690
And here we notice that terms
in the numerator and

00:05:16.690 --> 00:05:23.350
denominator cancel out, and we
obtain 8 choose 1 divided by

00:05:23.350 --> 00:05:25.690
10 choose 3.

00:05:25.690 --> 00:05:29.480
And to simplify things
just a little more,

00:05:29.480 --> 00:05:31.090
what is 8 choose 1?

00:05:31.090 --> 00:05:34.150
This is the number of ways that
we can choose one item

00:05:34.150 --> 00:05:36.090
out of eight items.

00:05:36.090 --> 00:05:37.680
And this is just 8.

00:05:42.220 --> 00:05:45.930
And let's leave the denominator
the way it is.

00:05:45.930 --> 00:05:51.150
So this is the answer to the
question that we had.

00:05:51.150 --> 00:05:55.840
And now let us work towards
developing a second approach

00:05:55.840 --> 00:05:58.140
towards this particular
answer.

00:06:00.710 --> 00:06:05.790
In our second approach, we start
first by looking at the

00:06:05.790 --> 00:06:08.750
sample space and understanding
what

00:06:08.750 --> 00:06:12.110
conditioning is all about.

00:06:12.110 --> 00:06:16.760
In our model, we have
a sample space.

00:06:16.760 --> 00:06:19.660
As usual we can denote
it by omega.

00:06:19.660 --> 00:06:25.380
And the sample space contains a
bunch of possible outcomes.

00:06:25.380 --> 00:06:33.090
A typical outcome is going to
be a sequence of length 10.

00:06:33.090 --> 00:06:36.409
It's a sequence of
heads or tails.

00:06:36.409 --> 00:06:39.070
And it's a sequence that
has length 10.

00:06:42.680 --> 00:06:45.870
We want to calculate conditional
probabilities.

00:06:45.870 --> 00:06:50.290
And this places us in a
conditional universe.

00:06:50.290 --> 00:06:54.500
We have the conditioning event
B, which is some set.

00:06:58.280 --> 00:07:02.010
And conditional probabilities
are probabilities defined

00:07:02.010 --> 00:07:05.810
inside this set B and define
the probabilities, the

00:07:05.810 --> 00:07:09.140
conditional probabilities of
the different outcomes.

00:07:09.140 --> 00:07:11.460
What are the elements
of the set B?

00:07:11.460 --> 00:07:17.140
A typical element of the set B
is a sequence, which is, again

00:07:17.140 --> 00:07:23.630
of length 10, but has
exactly three heads.

00:07:23.630 --> 00:07:26.593
So these are the three-head
sequences.

00:07:32.580 --> 00:07:37.750
Now, since we're conditioning
on event B, we can just work

00:07:37.750 --> 00:07:39.790
with conditional
probabilities.

00:07:39.790 --> 00:07:44.850
So let us find the conditional
probability law.

00:07:44.850 --> 00:07:50.880
Recall that any three-head
sequence has the same

00:07:50.880 --> 00:07:55.520
probability of occurring in
the original unconditional

00:07:55.520 --> 00:07:59.870
probability model, namely as
we discussed earlier, any

00:07:59.870 --> 00:08:04.560
particular three-head sequence
has a probability equal to

00:08:04.560 --> 00:08:06.530
this expression.

00:08:06.530 --> 00:08:09.150
So three-head sequences are
all equally likely.

00:08:09.150 --> 00:08:11.740
This means that the
unconditional probabilities of

00:08:11.740 --> 00:08:14.790
all the elements of
B are the same.

00:08:14.790 --> 00:08:18.300
When we construct conditional
probabilities given an event

00:08:18.300 --> 00:08:23.510
B, what happens is that the
ratio or the relative

00:08:23.510 --> 00:08:28.520
proportions of the probabilities
remain the same.

00:08:28.520 --> 00:08:31.640
So conditional probabilities
are proportional to

00:08:31.640 --> 00:08:34.070
unconditional probabilities.

00:08:34.070 --> 00:08:36.730
These elements of B were
equally likely in

00:08:36.730 --> 00:08:38.159
the original model.

00:08:38.159 --> 00:08:41.960
Therefore, they remain equally
likely in the conditional

00:08:41.960 --> 00:08:43.870
model as well.

00:08:43.870 --> 00:08:48.260
What this means is that the
conditional probability law on

00:08:48.260 --> 00:08:50.910
the set B is uniform.

00:08:50.910 --> 00:08:55.140
Given that B occurred, all the
possible outcomes now have the

00:08:55.140 --> 00:08:56.910
same probability.

00:08:56.910 --> 00:08:59.310
Since we have a uniform
probability law, this means

00:08:59.310 --> 00:09:01.560
that we can now answer
probability

00:09:01.560 --> 00:09:03.940
questions by just counting.

00:09:03.940 --> 00:09:06.850
We're interested in the
probability of a certain

00:09:06.850 --> 00:09:11.640
event, A, given that
B occurred.

00:09:11.640 --> 00:09:15.660
Now, given that B occurred, this
part of A cannot happen.

00:09:15.660 --> 00:09:18.900
So we're interested in the
probability of outcomes that

00:09:18.900 --> 00:09:23.240
belong in this shaded region,
those outcomes that belong

00:09:23.240 --> 00:09:28.830
within the set B. To find the
probability of this shaded

00:09:28.830 --> 00:09:33.080
region occurring, we just need
to count how many outcomes

00:09:33.080 --> 00:09:37.220
belong to the shaded region and
divide them by the number

00:09:37.220 --> 00:09:41.290
of outcomes that belong
to the set B.

00:09:41.290 --> 00:09:44.880
That is, we work inside this
conditional universe.

00:09:44.880 --> 00:09:47.230
All of the elements in this
conditional universe are

00:09:47.230 --> 00:09:48.700
equally likely.

00:09:48.700 --> 00:09:51.860
And therefore, we calculate
probabilities by counting.

00:09:51.860 --> 00:09:55.410
So the desired probability is
going to be the number of

00:09:55.410 --> 00:10:00.520
elements in the shaded region,
which is the intersection of A

00:10:00.520 --> 00:10:08.280
with B, divided by the number of
elements that belong to the

00:10:08.280 --> 00:10:13.000
set B.

00:10:13.000 --> 00:10:18.270
How many elements are there in
the intersection of A and B?

00:10:18.270 --> 00:10:22.590
These are the outcomes or
sequences of length 10, in

00:10:22.590 --> 00:10:24.980
which the first two tosses
were heads--

00:10:24.980 --> 00:10:26.940
no choice here.

00:10:26.940 --> 00:10:29.630
And there is one more head.

00:10:29.630 --> 00:10:33.350
That additional head can appear
in one out of eight

00:10:33.350 --> 00:10:34.830
possible places.

00:10:34.830 --> 00:10:38.240
So there's eight possible
sequences that have the

00:10:38.240 --> 00:10:40.710
desired property.

00:10:40.710 --> 00:10:44.010
How many elements are
there in the set B?

00:10:44.010 --> 00:10:49.270
How many three-head sequences
are there?

00:10:49.270 --> 00:10:54.800
Well, the number of three-head
sequences is the same as the

00:10:54.800 --> 00:10:58.920
number of ways that we can
choose three elements out of a

00:10:58.920 --> 00:11:01.450
set of cardinality 10.

00:11:01.450 --> 00:11:07.580
And this is 10 choose 3, as
we also discussed earlier.

00:11:07.580 --> 00:11:11.750
So this is the same answer as
we derived before with our

00:11:11.750 --> 00:11:13.100
first approach.

00:11:13.100 --> 00:11:17.650
So both approaches, of course,
give the same solution.

00:11:17.650 --> 00:11:21.310
This second approach is a little
easier, because we

00:11:21.310 --> 00:11:24.730
never had to involve any
p's in our calculation.

00:11:24.730 --> 00:11:27.210
We go to the answer directly.

00:11:27.210 --> 00:11:31.110
The reason that this approach
worked was that the

00:11:31.110 --> 00:11:35.970
conditional universe, the
event B, had a uniform

00:11:35.970 --> 00:11:37.300
probability law on it.