You are on page 1of 4

Solution of Airplane Problem, Lie detector problem and some

more practice questions

14 August 2019

1 Solution to Airplane Problem


We will be working with the rule that same policy will apply for all airplanes. For example, if our
policy says that the number of tickets to be sold is 182, it would apply for all the 200 airplanes.
We ultimately need to find out the probability of facing a embarrassing day. We would break this
task into 2 parts. We first compute the probability of a single airplane to face a embarrassing sit-
uation. In the second part, we would be using this probability to compute the probability of not
facing a embarrassing day.

Let us consider only one airplane. Denote by N the number of tickets to be sold. Denote by X
the random variable, which would denote the number of customers who would be turning up.
Please note that X is the random variable, N is not the random variable. N is a decision variable
which we need to determine.

For this airplane, we would be interested in finding out the probability of facing an embarrassing
situation.
Is is given that about 3% of the customers don’t turn up. Thus, it can be interpreted as the proba-
bility of a person not turning up.
Hence, Probability of a person turning up is equal to 0.97.
Since, the same rule applies to all individuals, it can be implied that X is a binomial random variable
with parameters N and 0.97. If you can visualize this as a coin toss experiment with the probability
of heads as 0.97, and number of tosses as N , it would be helpful in deciphering that X will follow a
binomial distribution.
(Please convince yourself with this, this is the most important step, from this point, it is only a
matter of calculations. Also, one can always argue that in airplanes, the trials are not independent,
as people might be booking in groups and also possibly cancelling in groups. However, when we say
it is binomial random variable, we are making the assumption that all trials are independent. In
order to proceed further, we will work with the assumption that trials are independent.)
The airline will not face an embarrassing situation when there are less than 180 passengers who turn
up.
Let pairplane = P (X ≤ 180), which can be computed using BINOMDIST function in excel, with
number of success = 180, Number of trials = N (you need to experiment with N ), probability of
success = 0.97, and since you want cumulative, you should give the last parameter as TRUE.
We are now done with the first part, and now need to find the probability of not facing a em-
barrassing day. The airline will not face an embarrassing day when none of the airplane faces an

1
embarrassing situation. Since the airplanes are operating independently of each other, this will be
given by p200
airplane . There are 2 ways of arriving at this -

• Define Ei as the event as ith airplane doesn’t face a embarrassing situation. You are interested
in the event none of the airlines face embarrassing situation, which is the event E1 ∩ E2 ∩ ... ∩
E200 .
P (E1 ∩ E2 ∩ ... ∩ E200 = P (E1 )P (E2 )...P (E200 ), as all these events are independent of each
other.
• Once you have obtained pairplane . One can model this again as a binomial random variable with
200 trials and probability of success as pairplane . We are now interested in finding the prob-
ability that this random variable takes the value 200, which can be done using BINOMDIST
function in excel(the fourth parameter now would be FALSE)
You might be surprised that even with N = 181, the probability that the airline doesn’t faces an
embarrassing day comes to be as low as 0.45 approximately.

2 Lie detector machine - Conditional Probability


We will use the following abbreviations -
1. PL = Person is Liar

2. PH = Person is Honest
3. MSH = Machine says the person is honest
4. MSL = Machine says the person is Liar

We first list down what is given about the machine -

1. P (M SH|P H) = 0.98
2. Hence, P (M SL|P H) = 0.02

3. P (M SL|P L) = 0.99
4. Hence, P (M SH|P L) = 0.01
Also, it is given that P (P H) = 0.97, Hence P (P L) = 0.03.

It is now given that machine reports a person to be liar. We are required to find
the probability that the person is honest?
Thus, we have to find P (P H|M SL). Note that this is not a direct question, this is like a inference
we have to make after the observation that machine has reported that a person is a liar.
Before we get into the conditional probabilities and related formulae, let us convince ourself with
assuming some numbers, possible large enough, and try to understand the interprtation of the given
probabilities.
Assume 1,00,000 people in the population. Since, 3% are liars, number of liars = 3000, number of
honest people is 97,000. The value P (M SH|P H) = 0.98 implies that if honest people are subjected

2
to the test by the machine, 98% of the times, machine will report them as honest. Hence, if all 97,000
are subjected to this test, 1940 will be reported as liars and the remaining 95060 will be reported
correctly as honest. Similarly, one can interpret P (M SL|P L) = 0.99. We thus, have the following
table. From this table, one can try to find the answer now. Since, we are given that machine says

Total Honest Liars


100000 97000 3000
Machine says Honest 95060 30
Machine says Liars 1940 2970

liar. The person is someone from the second row of the table. The probability of the person being
1940
honest given that the machine has reported him to be liar, hence, is 1940+2970 .

Now, we will use conditional probability to get the answer.


There are three important formulae that we shall use in the computations -

P (A ∩ B)
1. For any 2 events A and B, P (A|B) = .
P (B)
P (B ∩ A)
Since, this is true for any 2 events, it can also be said that P (B|A) =
P (A)
Hence,
2. P (A|B)P (B) = P (A ∩ B) = P (B|A)P (A)
because P (A ∩ B) = P (B ∩ A)
3. Law of total probability -
Consider any event C. If you can find events A1 , A2 , ..., An such that all of them mutually
exclusive and exhaustive (union of those events is the entire sample space), it can be said that
P (C) = P (C ∩ A1 ) + P (C ∩ A2 ) + ... + P (C ∩ An )
or alternatively,
P (C) = P (C|A1 )P (A1 ) + P (C|A2 )P (A2 ) + ... + P (C|An )P (An )

We now will apply these results to obtain the solution to P (P H|M SL). Note that the objective
would be to express this in terms of the probabilities that are known.
P (P H ∩ M SL) P (M SL|P H)P (P H)
Hence, P (P H|M SL) = =
P (M SL) P (M SL)
Note that the numerator is now known, to compute the denominator, we use the law of total proba-
bility by dividing the sample space into two events, Person is honest(PH) and Person is Liar(PL).
Note that these events are mutually exclusive and exhaustive. Hence,
P (M SL|P H)P (P H) P (M SL|P H)P (P H)
P (P H|M SL) = =
P (M SL) P (M SL|P H)P (P H) + P (M SL|P L)P (P L)
Now, you have expressed everthing in terms of the known quantities. Do the calculations and verify
that it matches with the one computed above.

3
3 Practice Problems
1. Let X be a random variable that follows a normal distribution with mean = 150 and variance
= 64. Using normdist function in excel, compute the following -
(a) P (X ≤ 160)
(b) P (140 ≤ X ≤ 160)
(c) P (X ≥ 140)
(d) The density of the function at X = 150, X = 140 and X = 160. Note that since X is a
continuous random variable, P (X = 150) = P (X = 160) = P (X = 140) = 0.
2. Let X be a random variable that follows a normal distribution with mean = µ and variance
= σ2 .
X −µ
Consider another random variable Z =
σ
Compute the mean and variance of Z.
NOTE - You should get this as 0 and 1. A normal distribution that has mean 0 and variance
1 is called as standard normal distribution. Thus, this is a method to convert any normal
distribution to a standard normal distribution.

3. Let X follow a uniform distribution with minimum value as 4 and maximum value as 10. What
is the probability that -
(a) X is between 5 and 9
(b) X is lesser than 5
(c) X is greater than 8
(d) X is greater than 9 or lesser than 5
Repeat problems (a)-(d) above if X is a random variable that follow uniform distribution but
the minimum value is -4 and maximum value is 10.

4. If the probability of hitting a target is 0.4 and 10 shots are fired independently, what is the
probability of target being hit atleast thrice?
5. In the above problem, find the conditional probability of the target being hit atleast thrice
given that atleast one hit is scored?
HINT : Define event A as target being hit atleast thrice, event B as target being hit atleast
once. You are asked to find P (A|B). Note that A ∩ B = A
6. Experience shows that 20% of the people reserving tables at a certain restaurant don’t show up.
If the restaurant has 40 tables and it takes reservation for 45 people, what is the probability
it will be able to accomodate everyone?

You might also like