MATH F113: Probability & Statistics

Second Semester 2024-2025

Divyum Sharma
Assistant Professor
Department of Mathematics
BITS Pilani
Lecture 11 Summary

1. Geometric distribution
2. Poisson distribution, Poisson approximation to
binomial distribution

Probability & Statistics Divyum Sharma 1

Recall: Poisson Distribution

Definition
A discrete random variable X is said to have a
Poisson distribution with parameter µ (µ > 0) if its
pmf is given by
µ x
e µ
p(x; µ) = , x = 0, 1, 2, . . .
x!

Probability & Statistics Divyum Sharma 2

Recall

Proposition (Mean & Variance)

If X has a Poisson distribution with parameter µ,
then E (X ) = V (X ) = µ.

Theorem
For a Poisson random variable X with parameter
µ > 0,
mX (t) = e µ(e 1) .
t

Probability & Statistics Divyum Sharma 3

Recall: The relationship between bino-
mial and Poisson rv

Let X be a binomial rv with pmf b(x; n, p). If

n ! 1, p ! 0, and np ! µ, then b(x; n, p) !
p(x; µ).

So in any binomial experiment in which n is large and

p is small such that np ! µ, we have
b(x; n, p) ⇡ p(x; µ). As a rule of thumb, this
approximation can safely be applied if n > 50 and
np < 5. This is also useful as binomial cdf tables
won’t be available for large n.
Probability & Statistics Divyum Sharma 4
Cdf of Poisson distribution

Cdf values are tabulated in Table A.2 (p. A-4 to A-5)

of the text book.

Probability & Statistics Divyum Sharma 5

Cdf of Poisson distribution

..
.

Probability & Statistics Divyum Sharma 6

Recall: (Poisson Process) We make the
following 3 assumptions about the way in which the
events of interest occur.

Probability & Statistics Divyum Sharma 7

Recall: (Poisson Process) We make the
following 3 assumptions about the way in which the
events of interest occur.
Time homogeneity and small interval probabilities:
1. There exists a parameter ↵ > 0 such that for any
short time interval of length t, the probability that
exactly one event occurs is ↵ · t + o( t). (Note
that o( t)/ t ! 0 as t ! 0.)

Probability & Statistics Divyum Sharma 7

Recall: (Poisson Process) We make the
following 3 assumptions about the way in which the
events of interest occur.
Time homogeneity and small interval probabilities:
1. There exists a parameter ↵ > 0 such that for any
short time interval of length t, the probability that
exactly one event occurs is ↵ · t + o( t). (Note
that o( t)/ t ! 0 as t ! 0.)
2. The probability of more than one event occurring
during any short time interval of length t is o( t).

Probability & Statistics Divyum Sharma 7

Recall: (Poisson Process) We make the
following 3 assumptions about the way in which the
events of interest occur.
Time homogeneity and small interval probabilities:
1. There exists a parameter ↵ > 0 such that for any
short time interval of length t, the probability that
exactly one event occurs is ↵ · t + o( t). (Note
that o( t)/ t ! 0 as t ! 0.)
2. The probability of more than one event occurring
during any short time interval of length t is o( t).
3. The number of events occurring during a short
time interval of length t is independent of the
number that occur prior.
Probability & Statistics Divyum Sharma 7
Let Pk (t) denote the probability that k events will be observed
during any particular time interval of length t (for
k = 0, 1, 2, . . .).

Probability & Statistics Divyum Sharma 8

Let Pk (t) denote the probability that k events will be observed
during any particular time interval of length t (for
k = 0, 1, 2, . . .).

↵t k
• Pk (t) = e k!(↵t) , so that the number of events
occurring during a time interval of length t is a
Poisson rv with parameter µ = ↵t.

• The expected number of events during a time

interval of length t is ↵t. So the expected number
during a unit interval of time is ↵.

The occurrence of events over time in this manner is called a

Poisson process and the parameter ↵ specifies the rate for
the process.

Probability & Statistics Divyum Sharma 8

Question
Hospital administrators in large cities anguish about
traffic in emergency rooms. At a particular hospital
in a large city, the staff on hand cannot
accommodate the patient traffic if there are more
than 10 emergency cases in a given hour. It is
assumed that patient arrival follows a Poisson
process, and historical data suggest that, on the
average, 5 emergencies arrive per hour.

(i) What is the probability that in a given hour the

staff cannot accommodate the patient traffic?
(ii) What is the probability that more than 20
emergencies arrive during a 3-hour shift?
Probability & Statistics Divyum Sharma 9
Solution

Let X = the number of emergency cases arriving in a

t hour period. As patient arrival follows a Poisson
process, X is a Poisson rv with parameter µ = 5t.

(i) Here, t = 1. Hence

P(X > 10) = 1 F (10; 5) = 1 0.986 = 0.014
(ii) Since t = 3 (hours), we have µ = 5 ⇥ 3 = 15.
Hence P(X > 20) = 1 F (20; 15) =
1 0.917 = 0.083.

Probability & Statistics Divyum Sharma 10

Instead of observing events over time, consider
observing events of some type that occur in a two- or
three-dimensional region. For example, we might
select on a map a certain region R of a forest, go to
that region, and count the number of trees. Each
tree would represent an event occurring at a
particular point in space. Under assumptions similar
to 1–3, it can be shown that the number of events
occurring in a region R has a Poisson distribution
with parameter ↵ · a(R), where a(R) is the area of
R. The quantity ↵ is the expected number of events
per unit area or volume.
Probability & Statistics Divyum Sharma 11
Question
Coliform bacteria occur in river water with an
average of 1 bacteria per 10 cubic centimeters of
water.

1. Find the probability there are no bacteria in a 20

cubic centimeter sample of water which is
tested.
2. Find the probability there are 5 or more bacteria
in a 50 cubic centimeter sample.

Probability & Statistics Divyum Sharma 12

Solution. Let X = number of bacteria in a t cubic
centimeter sample of water. Since ↵ = 0.1, the pmf
of X is Poisson with µ = 0.1t.

1. Here, t = 20 and hence µ = 2. Therefore

e 2 20
P(X = 0) = = 0.135.
0!
2. Here, t = 50 and hence µ = 5. Therefore the
pmf of X at x equals e 5 5x /x! for
x = 0, 1, 2 . . .. Hence

P(X 5) = 1 P(X  4) = 0.560.

Probability & Statistics Divyum Sharma 13
§3.6, Q91

Probability & Statistics Divyum Sharma 14

(a) Let X denote the number of trees in a quarter
acre of plot. Then X has Poisson distribution
with parameter 80 ⇥ 0.25 = 20. Hence
P(X  16) = F (16; 20) = 0.221.
(b) The expected number of trees equals
80 ⇥ 85000 = 6800000.
(c) Since the area of the circle is
⇡r 2 = ⇡ ⇥ 0.01 ⇥ 640 = 20.106 acres, X has
Poisson distribution with parameter
80 ⇥ 20.106 = 1608.5. Write the pmf.

Probability & Statistics Divyum Sharma 15

Chebyshev’s Inequality

Informally, if the variance of X is small, then it is

unlikely to deviate very much from the mean. More
precisely,
Let X be a random variable with mean µ and finite
non-zero variance 2 . Then for any real number
k > 0, we have
1
P(|X µ| k ) .
k2

Probability & Statistics Divyum Sharma 16

 Chebyshev’s Inequality

Informally, if the variance of X is small, then it is

unlikely to deviate very much from the mean. More
precisely,
Let X be a random variable with mean µ and finite
non-zero variance 2 . Then for any real number
k > 0, we have
1
P(|X µ| k ) .
k2
Note. 1. If k  1, then 1/k 2 1 and hence the inequality
holds trivially.
Probability & Statistics Divyum Sharma 16
Question
Blogposts from a website have an average of 800
words per article with a standard deviation of 150
words. A blogpost is selected randomly from this
website. Use Chebyshev’s inequality to determine a
lower bound for the probability that it has between
500 and 1100 words.

Probability & Statistics Divyum Sharma 17

Solution. By Chebyshev’s inequality,
1
P(|X 800| 150k)  .
k2
Taking k = 2, we get
P(|X 800| 300)  1/4 = 0.25. Hence the
probability that the blogpost has between 500 and
1100 words is at least 0.75.

Probability & Statistics Divyum Sharma 18

Chebyshev’s inequality is sharp

Let k 1 and X be a random variable with pmf

p(x) given by
8
>
> 1/2k 2 if x = 1
>
>
>
<1 1/k 2 if x = 0
p(x) =
>
> 1/2k 2 if x = 1
>
>
>
:0 otherwise.

Then µ = 0 and = 1/k . Hence

P(|X µ| k ) = P(|X 0| 1)=1/k 2 .

Probability & Statistics Divyum Sharma 19
Example. If X denotes the number of heads in n
independent tosses of a fair coin, then µ = n/2 and
p
= n/2. Using Chebyshev’s inequality, we get that
the probability that X deviates from µ by more than
p
3 n is at most 1/36.

Probability & Statistics Divyum Sharma 20

Example. If X denotes the number of heads in n
independent tosses of a fair coin, then µ = n/2 and
p
= n/2. Using Chebyshev’s inequality, we get that
the probability that X deviates from µ by more than
p
3 n is at most 1/36.
Example. Suppose X is a Poisson rv with
parameter µ. Then
p 1
P(|X µ| 2 µ)  .
4

Probability & Statistics Divyum Sharma 20

Question
A car manufacturing company claims that for a
particular model of a car, the average time to a
major repair is 7 years with a standard deviation of
0.75 years. Assume that the car is purchased by
taking a loan, to be paid in 4 years. Use
Chebyshev’s inequality to determine an upper
bound for the probability that the car will undergo
a major repair before the loan is paid off.

Probability & Statistics Divyum Sharma 21

Solution. Let X denote the time to a major repair.
We have

P(X  4)  P(|X 7| 3) = P(|X 7| 4 ⇥ 0.75)

1 1
 2= = 0.0625. (by Chebyshev’s inequal
4 16
Hence the probability that the car will undergo a
major repair before the loan is paid off is at most
0.0625.

Probability & Statistics Divyum Sharma 22

Ch 4: Continuous Random
Variables & Probability
Distributions

Probability & Statistics Divyum Sharma 22

Recall: Two Types of Random Variables
Definition
A random variable is said to be discrete if its
possible values either constitute a finite set or a
countably infinite set. A random variable X is said
to be continuous if both of the following hold.

1. The set of possible values of X is either an

interval or a disjoint union of intervals.
2. P(X = c) = 0 for any possible value of c (i.e.,
no possible value of X has a positive
probability.)
Probability & Statistics Divyum Sharma 23
Definition. Let X be a continuous random vari-
able. Then a probability distribution or probability
density function (pdf) of X is a function f such
that for any two numbers a, b with a  b, we
have Z b
P(a  X  b) = f (x) dx.
a
The graph of f is called the density curve.

Probability & Statistics Divyum Sharma 24

Probability & Statistics Divyum Sharma 25
 Definition. Let X be a continuous random vari-
able. Then a probability distribution or probability
density function (pdf) of X is a function f such
that for any two numbers a, b with a  b, we
have Z b
P(a  X  b) = f (x) dx.
a
The graph of f is called the density curve.

For f to be a legitimate pdf, it must satisfy the

following two conditions.
(i) f (x) 0 for all x
R1
(ii) 1 f (x) dx = 1
Probability & Statistics Divyum Sharma 26
Let X be a continuous rv with pdf f .
1. f (x) 6= P(X = x).
2. For any x0 , we have
Z x0
P(X = x0 ) = P(x0  X  x0 ) = f (x) dx = 0.
x0

3. For any finite set A, we have

P(A) = P(x 2 A) = 0.

4. For any two numbers a, b with a  b, we have

P(a  X  b) = P(a < X  b)

= P(a  X < b) = P(a < X < b).
Probability & Statistics Divyum Sharma 27