MATH 22

ENGINEERING DATA ANALYSIS

© https://uccs.edu/dase/

MODULE 3: DISCRETE PROBABILITY DISTRIBUTIONS

For educational purposes only

This module has the following sections and corresponding icons:

COVER PAGE
It shows the course code and title and
the specific topic discussed in this
module.

OVERVIEW
A general introduction of the specific topic
presented in the title, as well as the learning
outcomes and the time frame is included in this
section.

LECTURE
This section focuses on the discussion of the topic
which may include graphs, tables, and pictures to
help the student/reader in understanding the lesson.

PRACTICE PROBLEMS
It involves questions or follow-up problems about the
topic to evaluate how much you understand in this
module.
 ASSESSEMENT
To further assess your understanding, this section
will contain more challenging problems and real-life
situations in which you can apply what you have
learned throughout this module

SUPPLEMENTARY KNOWLEDGE
This section will help you in exploring the lesson by
providing tips, techniques, and other relevant information
to support your knowledge.

ANSWER KEY
This contains answers to all the problems included in
this module.

REFERENCES
This includes the list of all the reference books used
in creating this module.
 Overview
INTRODUCTION:
One engineering firm enjoys 40% success rate in getting state
government construction contracts. This month they have submitted bids on
eight construction projects to be funded by the state government. How likely is
this firm to get none of those contracts? Five out of eight contracts? All eight
contracts? To answer such questions, we obtain the probability distribution
that describes the likelihood of all possible outcomes in a given situation.
Whether an experiment yields qualitative or quantitative outcomes,
methods of statistical analysis require that we focus on certain numerical
aspects of the data (such as a sample proportion x/n, mean, or standard
deviations). The concept of a random variable allows us to pass from the
experimental outcomes themselves to a numerical function of the outcomes.
There are two fundamentally different types of random variables—discrete
random variables and continuous random variables. In this module, we
examine the basic properties and discuss the most important examples of
discrete variables.

COURSE OUTCOMES:
At the end of this module, students will be able to:
Apply statistical methods in the analysis of data and designing experiments.

INTENDED LEARNING OUTCOMES:

At the end of this module, the students will be able to:
Calculate probabilities, means, and variances for some common discrete
probability distributions

PROGRAM OUTCOMES:
Design a system, component, or process to meet desired needs within
realistic constraints such as economic, environmental, social, political,
ethical, health and safety, manufacturability, and sustainability, in
accordance with standards;
Identify, formulate, and solve complex problems in electrical engineering;
and
Apply techniques, skills and modern engineering tools necessary for
electrical engineering practice.

TIME FRAME:
 1 week
 LECTURE 1:
RANDOM VARIABLES AND THEIR
PROBABILITY DISTRIBUTIONS

In any experiment, there are numerous characteristics that can be observed or measured,
but in most cases an experimenter will focus on some specific aspect or aspects of a sample.
For example, in a study of commuting patterns in a metropolitan area, each individual in a
sample might be asked about commuting distance and the number of people commuting in
the same vehicle, but not about IQ, income, family size, and other such characteristics.
Alternatively, a researcher may test a sample of components and record only the number
that have failed within 1000 hours, rather than record the individual failure times. In general,
each outcome of an experiment can be associated with a number by specifying a rule of
association (e.g., the number among the sample of ten components that fail to last 1000
hours or the total weight of baggage for a sample of 25 airline passengers). Such a rule of
association is called a random variable—a variable because different numerical
values are possible and random because the observed value depends on which of the
possible experimental outcomes results (Figure 3.1).

Figure 3.1 A random variable

Adapted from Probability and Statistics for Engineering and the Sciences by Jay L. Devore, 8th Edition

Random variables are typically denoted by uppercase letters, such as X, Y, and Z. The
actual numerical values that a random variable can assume are denoted by lowercase
letters, such as x, y, and z. The probability that the random variable X takes a value x is
denoted by P(X = x) or P(x). We encounter two types of random variables, discrete and
continuous. In this module, we’ll study discrete random variables and their probability
distributions. In the next module, we’ll study continuous random variables and their
probability distributions.
Example 3.1
A section of an electrical circuit has two relays, numbered 1 and 2, operating in parallel. The
current will flow when a switch is thrown if either one or both of the relays close. The
probability of a relay closing properly is 0.8 for each relay. We assume that the relays
operate independently. Let Ei denote the event that relay i closes properly when the switch is
thrown. Then � �� = 0.8. A numerical event of some interest to the operator of this system
is
�� = the number of relays that close properly when the switch is thrown
Now X can take on only three possible values, 0, 1, and 2. We can find the probabilities
associated with these values of X by relating them to the underlying events �� .Thus, using
the rules in the past module, we get

� � = 0 = � �1 ∩ �2 = � �1 � �2 = 0.2 0.2 = 0.04

 Because X = 0 means neither relay closes when the relays operate independently.
Similarly,

� � = 1 = � �1 �2 ∪ �1 �2 = � �1 � �2 + � �1 � �2 = 0.8 0.2 + 0.2 (0.8) = 0.32

And

� � = 2 = � �1 ∩ �2 = � �1 � �2 = 0.8 0.8 = 0.04

The values of the random variable X along with their probabilities are useful for keeping track
of the operation of this system. The number of properly closing relays is the key to whether
or not the system will work. The current will flow if X is at least 1, and this event has
probability
� � ≥ 1 = � � = 1 + � � = 2 = 0.32 + 0.64 = 0.96
Note that we have mapped the outcomes of observing a process into a set of three
meaningful real numbers and have attached a probability to each.

A random variable X is said to be discrete if it can take on only a finite number, or a

countably infinite number, of possible values x.
Following are some of the examples of a discrete random variable:
• Number of defective diodes in a sample of 64
• Number of switches in “On” position, out of 20 switches on a circuit board
• Number of non-confirming assemblies out of 10
• Number of rolls of paper produced per day by a paper mill
• Number of defects per yard recorded for cable inspected
• Number of days recorded between consecutive failures of a network
In the relay example described earlier, the random variable X has only 3 possible
values, and it is a relatively simple matter to assign probabilities to these values.
The function p(x) is called the probability function of X, provided

1. � � = � = � � ≥ 0 for all values of �.

2. � � � = � = 1

Table 3.1 Probability distribution of number of closed Figure 3.2 Probability distribution of number of closed
relays relays

Adapted from Probability and Statistics for Engineers by Adapted from Probability and Statistics for Engineers by
Scheaffer, Mulekar, McClave, 5th ed. Scheaffer, Mulekar, McClave, 5th ed.
 The probability function is sometimes called the probability mass function of X to
denote the idea that a mass of probability is piled up at discrete points. It is often convenient
to list the probabilities for a discrete random variable in a table or plot in a chart. With X
defined as the number of closed relays, as in Example 3.1, the table and the graph of the
probability mass function are given in Table 3.1 and Figure 3.2, respectively. The listing and
the graph are two ways of representing the probability distribution of X.
Note that the probability function p(x) satisfies two requirements:
1. 0 ≤ � � ≤ 1 for � = 0, 1, 2
2. � 0 + � 1 + � 2 = 1

Example 3.2
The output of circuit boards from two assembly lines set up to produce identical
boards is mixed into one storage tray. As inspectors examine the boards randomly, it
is difficult to determine whether a board comes from line A or line Bo. A probabilistic
assessment of this situation is often helpful. Suppose a storage tray contains 10
circuit boards, of which 6 come from line A and 4 from line B. An inspector selects 2
of these identical-appearing boards for inspection. He is interested in X, the number
of inspected boards from line A. Find the probability distribution for X.
Solution:
The experimenter consists of two selections, each of which can result in two outcomes. Let
�� denote the event that the ith board comes from line A, and �� denote the event that it
comes from line B. Then the probability of selecting two boards from line is

� � = 2 = � �1 �2 = � � on first � � on second � on first)

The multiplicative rule of probability is used, and the probability of the second selection
depends on what happened on the first selection. There are other possibilities for outcomes
that will result in different values of X. These outcomes are conveniently listed using a tree
diagram in Figure 3.3. The probabilities for the various selections are given on the branches
of the tree. It is easily seen that X has three possible outcomes, with the probabilities listed
in Table 3.2.

Table 3.2
Figure 3.3 Outcomes of circuit board selection
Adapted from Probability and
Adapted from Probability and Statistics for Engineers by
Statistics for Engineers by Scheaffer,
Scheaffer, Mulekar, McClave, 5th ed.
Mulekar, McClave, 5th ed.
 PRACTICE PROBLEMS

3.1P Among 10 applicants for an open position, 6 are females and 4 are males.
Suppose three applicants are randomly selected from the applicant pool for final
interviews. Find the probability distribution for X, the number of female applicants
among the final three.

2.2P Wade Boggs of the Boston Red Sox hit 0.363 in 1987. (He got a hit on 36.3%
of his official times at bat.) In a typical game, he was up to bat three official times.
Find the probability distribution for X, the number of hits in a typical game. What
assumptions are involved in the answer? Are the assumptions reasonable? Is it
unusual for a good hitter like Wade Boggs to go zero for three in one game?
 LECTURE 2:
CUMULATIVE DISTRUBUTION
FUNCTIONS

Cumulative probabilities, �(� ≤ �), where X still represents the random variable and x now
represents an upper limit, are found by adding individual probabilities.

� �≤� = �(�� )
��≤�

Where �(�� ) is an individual probability function. For example, if �� can only be zero or
positive integer,
� �≤ 3 =� 0 +� 1 +� 2 +� 3
The functional relationship between the cumulative probability and the upper limit, x, is called
the cumulative distribution function, or the probability distribution function.
Note that since � � ≤ 2 = � 0 + � 1 + � 2 , we have � 3 = � � ≤ 3 − �(� ≤ 2)
In general,
� �� = � � ≤ �� − �(� ≤ ��−1 )
As an illustration, consider the random variable that represents the number of heads
obtained on tossing five fair coins. The probability of obtaining heads on any 1 one coin is 2.
The probability function and cumulative distribution are given by the binomial distribution,
which will be considered in detail in the next lecture. The probability function of possible
results is shown in Table 3.3 and Figure 3.5.

Figure 3.5 Probability Function for Results of

Table 3.3 Probability function for tossing coins Tossing Five Fair Coins
Adapted from Statistics and Probability for Adapted from Statistics and Probability for
Engineering Applications With Microsoft® Excel Engineering Applications With Microsoft® Excel
by W.J. DeCoursey, 2003 by W.J. DeCoursey, 2003
The corresponding cumulative distribution function is shown in Figure 3.6. The graph of the
cumulative distribution function for a discrete random variable is a stepped function because
there can be no change in the cumulative probability between possible values of the variable.
Using this cumulative distribution function with equation � �� = � � ≤ �� − �(� ≤ ��−1 )
26 16 10
� 3 =� �≤3 −� �≤2 = − = = 0.3125
32 32 32

Figure 3.5 Cumulative Distribution for Tossing Five Fair Coins

Adapted from Statistics and Probability for Engineering Applications with Microsoft® Excel by W.J. DeCoursey, 2003

Example 3.3
The random variable X, denoting the number of relays closing properly (defined in
Example 3.1) has the probability distribution given below:

P(X=0) P(X=1) P(X=2)

0.04 0.32 0.64

Note that � � ≤ 1.5 = � � = 0 + � � = 1 = 0.04 + 0.32 = 0.36

The distribution function for this random variable (graphed in Figure 3.4) has the form
 Figure 3.6 A distribution function for a discrete random variable
Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.

PRACTICE PROBLEMS

Verify that the following functions are cumulative distribution functions, and determine the
probability mass function and the requested probabilities.

0 �<1
3.3P � � = 0.5 1 ≤ � < 3
1 3≤�

a. � � ≤ 3 b. � � ≤ 2
c. � 1 ≤ � ≤ 2 d. � � > 3

3.4P

a. � � ≤ 50 b. � � ≤ 40
c. � 40 ≤ � ≤ 60 d. � � < 0
e. � 0 ≤ � < 10 f. � −10 < � < 10
 LECTURE 3:
EXPECTED VALUES OF RANDOM
VARIABLES

Two numbers are often used to summarize a probability distribution for a random variable X.
The mean is a measure of the center or middle of the probability distribution, and the
variance is a measure of the dispersion, or variability in the distribution. These two
measures do not uniquely identify a probability distribution. That is, two different distributions
can have the same mean and variance. Still, these measures are simple, useful summaries
of the probability distribution of X.
The mean or expected value of the discrete random variable X, denoted as μ or E(X) is,

�=� � = ��(�)
�

The variance of X, denoted as σ or V(X), is

2

�2 = � � = �(� − �)2 = � − � 2� � = �2 � � − (�)2

� �

The standard deviation of X is

�= �2
The mean of a discrete random variable X is a weighted average of the possible values of X
with weights equal to the probabilities. If f(x) is the probability mass function of a loading on
a long, thin beam, E(X) is the point at which the beam balances. Consequently, E(X)
describes the “center” of the distribution of X in a manner similar to the balance point of a
loading. See Fig. 3-5.

The variance of a random variable X is a measure of dispersion or scatter in the possible

values for X. The variance of X uses weight f(x) as the multiplier of each possible squared
deviation (x−μ)2. Figure 3.7 illustrates probability distributions with equal means but different
variances. Properties of summations and the definition of μ can be used to show the equality
of the formulas for variance.

� � = � − � 2� � = �2 � � − 2� �� � + �2 �(�)
� � � �

= �2 � � − 2�2 + �2 = �2 � � − (�)2
� �

Figure 3.7 A probability distribution can be viewed as a loading with the mean equal to the balance point.
Parts (a) and (b) illustrate equal means, but Part (a) illustrates a larger variance.
Adapted from Applied Statistics and Probability for Engineers by Montgomery and Runger, 5th ed.
 Figure 3.8 The probability distributions illustrated in Parts (a) and (b) differ even though they have equal
means and equal variances.
Adapted from Applied Statistics and Probability for Engineers by Montgomery and Runger, 5th ed.

Either formula for V (x) can be used. Figure 3.8 illustrates that two probability distributions
can differ even though they have identical means and variances.

Because probability can be thought of as the long-run relative frequency of occurrence for an
event, a probability distribution can be interpreted as showing the long-run relative frequency
of occurrence for numerical outcomes associated with a random variable.
Game 1: Suppose that you and your friend are matching balanced coins. Each of you
tosses a coin. If the upper faces match, you win $1; if they do not match, you
lose $1 (i.e., your friend wins $1).
The probability of a match is 0.5 and in the long run you will win about half the
time. Thus, a relative frequency distribution of your winnings should look like
Figure 3.9. Note that the negative sign indicates a loss to you.
On the average, how much will you win per game over the long run? If Figure
3.9 is a correct display of your winnings, you win -1 half the time and +1 half
the time, for an average of
1 1
−1 + +1 =0
2 2
This average is sometimes called your expected winnings per game or the
expected value of your winnings. An expected value of 0 indicates that this is
a fair game.
The expected value can be thought of as the mean value of your winnings
over many games.

Figure 3.9 Relative frequency of winnings

Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.
Game 2: Now payday has arrived, and you and your friend up the stakes to $10 per
game of matching coins. You now win -10 or +10 with equal probability. Your
expected winnings per game are
1 1
−10 + +10 =0
2 2
and the game is still fair.
The new stakes can be thought of as a function of the old in the sense that if
X represents winnings per game when you were playing for $1, then 10X
represents your winnings per game when you play for $10.

Game 3: You and your friend decide to play the coin–matching game by allowing you
to win $1 if the match is tails and $2 if the match is heads. You lose $1 if the
coins do not match. Notice that this is not a fair game, because your expected
winnings are
1 1 1
−1 + +1 + +2 = 0.25
2 4 4
So you are likely to win 25 cents per game in the long run. In other words, the
game is favorable to you.

Game 4: In the coin-matching game, suppose you pay your friend $1.50 if the coins do
not match. You’ll still win $1 if the match is tails and $2 if the match is heads.
Now your expected winnings per game are
1 1 1
−1.50 + +1 + +2 =0
2 4 4
and the game now is fair. What is the difference between Game 4 and the
original Game 1 in which the payoffs were $1?
The difference certainly cannot be explained by the expected value, because
both games are fair. You can win more but also lose more with the new
payoffs, so the difference between the two can be partially explained in terms
of the variation of your winnings across many games. This increased variation
can be seen in Figure 3.10, the relative frequency for your winnings in Game
4, which is more spread out than the one shown in Figure 3.3.

Figure 3.10 Relative frequency of winnings (Game 4)

Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.
By expanding the square term in the definition of variance and solving we get
� � = �(� − �)2 = �(�2 − 2�� + �2 )
= � �2 − 2�� � + �(�)2 = � �2 − �2
For the Game 1 depicted in Figure 3.9, the variance of your winnings (with � = 0) is
1 1
�2 = �(� − �)2 = −12 + +12 =1
2 2
It follows that as well. For the Game 4 depicted in Figure 3.10, the variance of your winnings
is
1 1 1
�2 = �(� − �)2 = (−1.52 ) + (+12 ) + (+22 ) = 2.375
2 4 4
And � = 2.375 = 1.54. Which game would you rather play? Game 1 or Game 4?
The standard deviation can be thought of as the size of a “typical” deviation between and
observed outcome and the expected value.
• In Figure 3.9, each outcome (-1 or +1) does deviate precisely one standard deviation from
the expected value.
• In Figure 3.10, the positive values deviate on the average 1.5 units from the expected
value of 0, as does the negative value, which is approximately one standard deviation below
the mean.
The mean and standard deviation provide a useful summary of the probability distribution for
a random variable that can assume many values.

Example 3.4
Table below provides the age distribution of the population (U.S. Bureau of Census)
for 1990 and 2050 (projected). The numbers are percentages.

Table 3.4
Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.
Age is actually a continuous measurement, but when reported in categories, we can treat it
as a discrete random variable. To move from continuous age intervals to discrete age
classes, we assign each interval the value of its midpoint (rounded). The data in table above
are interpreted as reporting that 7.6% of the 1990 population was around 3 years of age and
22.5% of the 2050 population is anticipated to be around 55 years of age. The open interval
at the upper end was stopped at 100 years for convenience.
With the percentages interpreted as probabilities, the mean age for 1990 is approximated by

�= �� � = 3 0.076 + 9 0.128 + … + 92 0.012 = 35.5

For 2050, the mean age is approximated by

�= �� � = 3 0.064 + 9 0.116 + … + 92 0.046 = 41.2

The mean age is anticipated to increase rather markedly. The variation in the two age
distributions can be approximated by the standard deviation. For 1990,

�= � − � 2� � = 3 − 35.5 2 0.076 + … + 92 − 35.5 2 0.012 = 25.5

and employing similar calculation for 2050 data, we get � = 25.4. These results are
summarized in table below.

Table 3.4
Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.

The population is not only getting older, on the average, but its variation is increasing too.
What are some of the implications of these trends?
 PRACTICE PROBLEMS

3.5P You are to pay $1 to play a game consisting of drawing one ticket at random from a
box of numbered tickets. You win the amount (in dollars) of the number on the ticket
you draw. Two boxes are available with numbered tickets as shown below:

3.6P The graph in Figure 3.11 shows the age distribution for AIDS deaths in the United
States through 1995. Approximate the mean and standard deviation of this age
distribution. How does the mean age compare to the approximate median age?

Figure 3.11 Distribution of AIDS deaths by age, 1982 through 1995

Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.
 LECTURE 4:
THE BINOMIAL DISTRIBUTION

Consider the following random experiments and random variables:

1. Flip a coin 10 times. Let � = number of heads obtained.
2. A worn machine tool produces 1% defective parts. Let � = number of defective
parts in the next 25 parts produced.
3. Each sample of air has a 10% chance of containing a particular rare molecule. Let
� = the number of air samples that contain the rare molecule in the next 18 samples
analyzed.
4. Of all bits transmitted through a digital transmission channel, 10% are received in
error. Let � = the number of bits in error in the next five bits transmitted.
5. A multiple-choice test contains 10 questions, each with four choices, and you
guess at each question. Let � = the number of questions answered correctly.
6. In the next 20 births at a hospital, let � = the number of female births.
7. Of all patients suffering a particular illness, 10% experience improvement from a
particular medication. In the next 100 patients administered the medication, let � =
the number of patients who experience improvement.
These examples illustrate that a general probability model that includes these experiments
as particular cases would be very useful.
Each of these random experiments can be thought of as consisting of a series of repeated,
random trials: 10 flips of the coin in experiment 1, the production of 25 parts in experiment 2,
and so forth. The random variable in each case is a count of the number of trials that meet a
specified criterion. The outcome from each trial either meets the criterion that � counts or it
does not; consequently, each trial can be summarized as resulting in either a success or a
failure. For example, in the multiple-choice experiment, for each question, only the choice
that is correct is considered a success. Choosing any one of the three incorrect choices
results in the trial being summarized as a failure.
The terms success and failure are just labels. We can just as well use A and B or 0 or 1.
Unfortunately, the usual labels can sometimes be misleading. In experiment 2, because X
counts defective parts, the production of a defective part is called a success.
A trial with only two possible outcomes is used so frequently as a building block of a random
experiment that it is called a Bernoulli trial. It is usually assumed that the trials that
constitute the random experiment are independent. This implies that the outcome from one
trial has no effect on the outcome to be obtained from any other trial. Furthermore, it is often
reasonable to assume that the probability of a success in each trial is constant. In the
multiple choice experiment, if the test taker has no knowledge of the material and just
guesses at each question, we might assume that the probability of a correct answer is ¼ for
each question
A random variable Y possesses a binomial distribution if
1. The experiment consists of a fixed number n of trials.
2. Each trial can result in one of only two possible outcomes, called “success” and “failure.”
3. The probability of “success,” p, is constant from trial to trial.
4. The trials are independent.
5. Y is defined as the number of successes among the n trials.

Many experimental and sample survey situations result in a random variable that can be
adequately modeled by the binomial distribution. For example,
• Number of defectives in a sample of n items from a large population
• Counts of number of employees favoring a certain retirement policy out of n employees
interviewed
• The number of pistons in an eight-cycle engine that are misfiring
• The number of electronic systems sold this week out of the n that are manufactured.

The binomial distribution is defined by two parameters, n (number of trials) and p (probability
of success on each trial). The binomial random variable Y = the number of successes in n
trials.
The Binomial Distribution
� �
� � = � (1 − �)�−� � = 0,1,2,3, …, � for 0 ≤ � ≤ 1
�
� � = �� and � � = ��(1 − �)

The figures on below show how the parameter values (n and p) affect the probability function
for the binomial distribution
 Figure 3.12 The effect of n, and p on the binomial probability function
Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.

From Figure 3.9 we see that

• For values of p near 0 or 1, the binomial distribution is skewed.
• As p gets closer to 0 (Figures 3.12a, and 3.12b), the binomial distribution
becomes more right-skewed, while as p gets closer to 1 (Figure 3.12d), the
binomial distribution becomes more left-skewed.
• For p = 0.5 (Figure 3.12c), the binomial distribution is symmetric and
mound-shaped.
• The distribution becomes more symmetric as n increases (Figure 3.12b and
3.12e).
The mean and variance of the binomial distribution can be determined using the basic
definition and solving the expression. This direct approach is tricky and requires regrouping
of the terms in the summation.
� �
� �
� � = �� � = � � (1 − �)�−� = ��
�
�=0 �=0

And
� �
� �
� � = � − � 2� � = (� − �)2 � (1 − �)�−� = ��(1 − �)
�
�=0 �=0

Example 3.5
Suppose a large lot contains 10% defective fuses. Four fuses are randomly sampled
from the lot.

(a) Find the probability that exactly one fuse in the sample of four is defective.
(b) Find the probability that at least one fuse in the sample of four is defective.
(c) Suppose the four fuses sampled from the lot were shipped to a customer
before being tested, on a guarantee basis. If any fuse is defective, the
 supplier will repair it without any change to the customer. Assume that the
cost of making the shipment good is C =3�2 , where Y denotes the number
of defectives in the shipment of four. Find the expected repair cost.
Solution:
We assume that the four fuses are sampled independently of each other and that the
probability of being defective is the same (0.1) for each fuse. This will be approximately true
if the lot is indeed large. If the lot is small, removal of one fuse would substantially change
the probability of observing a defective on the second draw. For a large lot, the binomial
distribution provides a reasonable model in this experiment with n = 4 and p = 0.1. Let
� = the number of defective fuses out of 4 inspected
(a) The probability that exactly one fuse in the sample of four is defective is

4
� �=1 =� 1 = 0.11 0.93 = 0.2916
1
(b) The probability that at least one fuse in the sample of four is defective is

4
� �≥1 =1−� �=0 =1− 0.10 0.94 = 0.3439
0

(c) We know that � � = � 3�2 = 3� � 2 . For the binomial distribution, � = � � =

� � and � � = �[(1 − �)

Because � � = �(� − �)2 = �(�)2 − �2 , we can get

�(�)2 = � � + �2 = �� 1 − � + (��)2

In this example, n = 4 and p = 0.1, and hence

� � = 3�(�)2 = 3 �� 1 − � + (��)2

= 3 4 0.1 0.9 + 42 (0.1)2 = 1.56

` If the cost were originally in tens of dollars, we could expect to pay an average of
$15.60 in repair costs for each shipment of four fuses.
 PRACTICE PROBLEMS

3.7P Let X denote a random variable having a binomial distribution with p = 0.2 and n = 4.
Find the following.
(a) P ( X = 2) (b) P ( X ≥ 2)
(c) P ( X ≤ 2) (d) E ( X )
(e) V ( X )

3.8P A machine that fills boxes of cereal underfills a certain proportion p.If 25 boxes are
randomly selected from the output of this machine, find the probability that no more
than two are underfilled when
(a) p = 0.1 (b) p = 0.2
 LECTURE 5:
THE POISSON DISTRIBUTION

The Poisson distribution occurs when we count the number of occurrences of an event over
a given time period or length or area or volume. For example:
• The number of flaws in a square yard of fabric
• The number of bacterial colonies in a cubic centimeter of water
• The number of times a machine fails in the course of a workday

The Poisson Distribution

�� �−�
� � = � = 0,1,2,3, …,
�!
� � =� and � � =�

A number of probability distributions come about through the application of limiting

arguments to other distributions. The Poisson distribution is one such distribution. It is
used as the limit of the binomial distribution.
Consider the development of a probabilistic model for the number of accidents occurring at a
particular highway intersection over a period of one week. We can think of the time interval
as being split up into n subintervals such that
� one accident in a subinterval = �
� no accidents in a subinterval = 1 − �
Note that we are assuming that the same value of p holds for all subintervals and that the
probability of more than one accident in any one subinterval is zero. If the occurrence of
accidents can be regarded as independent from one subinterval to another, then the total
number of accidents in the time period (which equals the total number of subintervals
containing one accident) will have a binomial distribution. There is no unique way to choose
the subintervals and we therefore know neither n nor p. Thus, we want to look at the limit of
the binomial probability distribution.
The Poisson distribution is defined by one parameter, namely � . The expected value of the
binomial distribution is np, where the expected value of the Poisson distribution is �. In order
to keep the mean in the binomial case constant at the Poisson mean (i.e., np= � ), it is
reasonable to assume that as n increases, p should decrease. With np= � , that is, p= �/�,
we have
Figure 3.13 shows how the choices of � affect the shape of a Poisson probability function.
Note that the probability function is very asymmetric for � = 1 (Figure 3.13a) but becomes
fairly symmetric and mound-shaped as � increases (Figures 3.13b and 3.13c).
We can intuitively determine what the mean and variance of a Poisson distribution should be
by recalling the mean and variance of a binomial distribution and the relationship between
the two distributions. A binomial distribution has mean np and variance �� 1 − � = �� −
��(�) . Now if n gets large and p remains at �� = � , the variance �� − �� � = � − ��
should tend toward � . In fact, the Poisson distribution does have both mean and variance
equal to � . The mean of the Poisson distribution is easily derived using the Taylor series
expansion of ex , namely,

�2 � 3
�� = 1 + � + + +…
2! 3!
Then,
 Figure 3.13 The Poisson Distribution
Adapted from Probability and Statistics for Engineers by Scheaffer, Mulekar,
McClave, 5th ed.

To find the variance, first find �(�)2 = � � � − 1 + � � . Then derive the variance using
similar arguments as in the expected value, and the relation � � = �(�)2 − [�(�)]2
The Poisson distribution applies in its own right where the possible number of discrete
occurrences is much larger than the average number of occurrences in a given interval of
time or space. The number of possible occurrences is often not known exactly. The
outcomes must occur randomly, that is, completely by chance, and the probability of
occurrence must not be affected by whether or not the outcomes occurred previously, so the
occurrences are independent. In many cases, although we can count the occurrences, such
as of a thunderstorm, we cannot count the corresponding non-occurrences. (We can’t count
“non-storms”!) Examples of occurrences to which the Poisson distribution often applies
include counts from a Geiger counter, collisions of cars at a specific intersection under
specific conditions, flaws in a casting, and telephone calls to a particular telephone or office
under particular conditions. For the Poisson distribution to apply to these outcomes, they
must occur randomly.

Example 3.6
For a certain Manufacturing industry, the number of industrial accidents averages
three per week.
(a) Find the probability that no accident will occur in a given week.
(b) Find the probability that two accidents will occur in a given week.
(c) Find the probability that at most four accidents will occur in a given week.
(d) Find the probability that two accidents will occur in a given day.

Solution:
(a) Using � = mean number of accidents per week = 3, we get
 30 −3
� no accident in a given week = � 0 = 0!
� = �−3 = 0.05

(b) Using � = mean number of accidents per week = 3, we get

32 −3
� Two accidents in a given week = � 2 = 2!
� = 0.224

(c) Using � = mean number of accidents per week = 3, we get

� at most 4 accidents in a given week = � 0 + � 1 + � 2 + � 3 + �(4)

30 −3 31 −3 32 −3 33 −3 34 −3
= � + � + � + � + � = 0.815
0! 1! 2! 3! 4!
(d) Now we are interested in the number of accidents on a given day. Thus,
using � = mean number of accidents per day = 3/7 = 0.2857, we get
0.28572 −0.2857
� Two accidents in a given week = � 2 = � = 0.031
2!

PRACTICE PROBLEMS

3.9P The number of telephone calls coming into the central switchboard of an office building
averages four per minute.
(a) Find the probability that no calls will arrive in a given 1-minute period.
(b) Find the probability that at least two calls will arrive in a given 1-minute period.
(c) Find the probability that at least two calls will arrive in a given 2-minute period.

3.10P The National Maximum Speed Limit (NMSL) of 55 miles per hour has been in force in
the United States since early 1974. The benefits of this law have been studied by D.B.
Kamerud (Transportation Research, 17A, no. 1, 1983, pp. 51–64), who reports that the
fatality rate for interstate highways with the NMSL in 1975 is approximately 16 per 109
vehicle miles
(a) Find the probability of at most 15 fatalities occurring in 109 vehicle miles.
(b) Find the probability of at least 20 fatalities occurring in 109 vehicle miles. (Assume
that the number of fatalities per vehicle mile follows a Poisson distribution.)
 ASSESSMENT

3.1A Forty percent of seeds from maize (modern-day corn) ears carry single spikelets, and
the other 60% carry paired spikelets. A seed with single spikelets will produce an ear
with single spikelets 29% of the time, whereas a seed with paired spikelets will
produce an ear with single spikelets 26% of the time. Consider randomly selecting
ten seeds. (20 points)
(a) What is the probability that exactly five of these seeds carry a single
spikelet and produce an ear with a single spikelet?
(b) What is the probability that exactly five of the ears produced by these
seeds have single spikelets? What is the probability that at most five ears
have single spikelets?

3.2A The manufacturer of a low-calorie dairy drink wishes to compare the taste appeal of a
new formula (B) with that of the standard formula (A). Each of four judges is given
three glasses in random order, two containing formula A and the other containing
formula B. Each judge is asked to state which glass he or she most enjoyed.
Suppose the two formulas are equally attractive. Let Y be the number of judges
stating a preference for the new formula. (30 points)
(a) Find the probability function for Y.
(b) What is the probability that at least three of the four judges state a
preference for the new formula?
(c) Find the expected value of Y.
(d) Find the variance of Y.

3.3A Of all customers purchasing automatic garage-door openers, 75% purchase a chain-
driven model. Let number among the next 15 purchasers who select the chain-driven
model. (30 points)
(a) What is the probability mass function (pmf) of �?
(b) Compute �(� > 10)
(c) Compute �(6 ≤ � ≤ 10)
(d) Compute � and �2 .
(e) If the store currently has in stock 10 chain-driven models and 8 shaft-
driven models, what is the probability that the requests of these 15 customers
can all be met from existing stock?

3.4A The probability that a single radar set will detect an airplane is 0.9. If we have five
radar sets, what is the probability that exactly four sets will detect the plane? At least
one set? (Assume that the sets operate independently of each other.) (20 points)
 SUPPLEMENTARY KNOWLEDGE

For additional information, you may view the following videos below:
1. Random Variables and their Probability Distributions
(https://www.youtube.com/watch?v=0P5WRKihQ4E)
2. Cumulative Distribution Functions (https://www.youtube.com/watch?v=3xAIWiTJCvE)
3. Expected Values of Random Variables
(https://www.youtube.com/watch?v=OvTEhNL96v0)
4. The Binomial Distribution (https://www.youtube.com/watch?v=qIzC1-9PwQo)
5. The Poisson Distribution (https://www.youtube.com/watch?v=jmqZG6roVqU)
6. Example problems (Binomial and Poisson with Geometric and Hypergeometric
(https://www.youtube.com/watch?v=Jm_Ch-iESBg)

ANSWER KEY
Answers to Practice Problems:

3.1P 3.2P

3.3P (a) 1 (b) 0.5 (c) 0.5 (d) 0.5

3.4P (a) 1 (b) 0.75 (c) 0.25 (d) 0.25 (e) 0 (f) 0
2 12
3.5P (a) 0, 3
(b) 0, 5
(c) Box II

3.6P (a) Mean = 37.17

(b) Standard Deviation = 11.45
(c) Median = 30-39

3.7P (a) 0.5136 (b) 0.1808 (c) 0.9728 (d) 0.8 (e) 0.64
3.8P (a) 0.537 (b) 0.098
3.9P (a) 0.0183 (b) 0.908 (c) 0.997

3.10P (a) 0.467 (b) 0.188

 Answers to Assessment Problems
3.1A (a) 0.0029 (b) 0.0767, 0.9702

4 1 2 1 4 8
3.2A (a) � � = �
(3 )�(3 )4−� (b) 9 (c) 3 (d) 9

3.3A (a) b(x; 15, 0.75) (b) 0.686 (c) 0.313 (d) 11.25, 2.81 (e) 0.310

3.4A 0.32805; 0.99999

References

1. R. Scheaffer, M. MuleKar and J. McClave, 2010, Probability and Statistics for

Engineering Students, 5th Edition
2. D. Montgomery and G. Runger, 2014, Applied Statistics and Probability for
Engineers, 6th Edition
3. W.J. DeCoursey, 2003, Statistics and Probability for Engineering Applications
With Microsoft® Excel
4. Jay L. Devore, 2010, Probability and Statistics for Engineering and the
Sciences, 8th Edition