BAYERO UNIVERSITY, KANO

FACULTY OF ENGINEERING
TEC4201: ENGINEERING STATISTICS
LECTURE NOTES BY Dr. A. A. ADAMU

1.0 INTRODUCTION – PROBABILITY

1.1 Basic Definitions

In statistics, we are basically concerned with the presentation and interpretation of chance
outcomes that occur in a planned study or scientific investigation. For example, the highway
engineer may record the number of accidents that occur monthly or the volume of traffic at Kabuga
junction, hoping to justify the installation of a traffic light; the electrical engineer would want to
know the frequency at which fuses blow or at which some IC’s burn out in order to determine the
safety (and hence reliability) of an electronic device; or the production engineer would want to
know the proportion of defective components coming from a manufacturing process to justify
switching to an alternative process of manufacture. We are therefore usually dealing with counts
or numerical measurements and the value so recorded is called observation. The numbers 3, 0, 2,
1, representing the number of accidents over a given time constitute a set of raw data. Any process
that generates raw data is called experiment.

1.2 Sample Space and Events

The set of all possible outcomes of a statistical experiment is called sample space, S. Each
outcome in a sample space is called an element or a member of the sample space or, simply, a
sample point.

Examples:
(i) Sample space for the experiment tossing a die once:
S = {1, 2, 3, 4, 5, 6}
- an example of a sample space with a finite number of elements.
(ii) Sample space for cities of the world with a population over 1 million:
S = { x/x is a city with population over 1 million}
- an example of sample space with a large or infinite number of sample space.
(iii) Sample space for three items selected at random from a manufacturing process and classified
as defective (D) or non-defective (N):
a) Most informative:
S = {NNN, NDN, DNN, NND, DDN, DND, NDD, DDD}
b) Least informative:
S = {0, 1, 2, 3},
Home work:
Review Venn diagrams and Counting sample points

1
1.3 Probability of an Event
The statistician is basically concerned with drawing conclusions or inferences from
experiments involving uncertainties. For these conclusion and inferences to be reasonably
accurate, an understanding of probability theory is essential.
To find the probability of any event A, we sum all the weights assigned to the sample points
in A. The sum is called the measure of A or the probability of A, denoted by P(A). Thus,
0 ≤ P(A) ≤ 1, P(φ) = 0, and P(S) = 1.
Hence, if an experiment can result in any one of N different equally likely outcomes, and if exactly
n of these outcomes corresponds to A, then the probability of event A is:
P ( A) =
n
N
Example 1: A coin is tossed twice. What is the probability that at least one head occurs?

Solution: The sample space is S = {HH, HT, TH, TT}

There are four points in the sample space and for an event A that at least a head occurs there are 3
points. Hence, P(A) = 3/4.

Example 2: A die is loaded in such a way that an even number is twice as likely to occur as an odd
number. If E is the event that a number less than 4 occurs on a single toss of the die, find P(E).

Solution: The sample space is S = {1, 2, 3, 4, 5, 6}. We assign a weight w to each odd number and
a weight 2w to each even number. Since the sum of the weights must be 1, we have 9w = 1 or w
= 1/9. Hence, weights of 1/9 and 2/9 are assigned to each odd number and even number,
respectively. Hence,
P (E ) = + + =
1 2 1 4
9 9 9 9

1.4 Conditional Probability

The probability of an event B occurring when it is known that some other event A has
occurred is called a conditional probability, and is denoted by P(B | A).

Example 1: What is the probability of getting a perfect square when a die is tossed, if the die is
constructed so that the even numbers are twice as likely to occur as the odd numbers? Suppose
now that it is known that the toss of the die resulted in a number greater than 3, find the probability
of getting a perfect square.

Solution: As in a previous example, we assign weights of 1/9 and 2/9 respectively to odd and even
numbers and if B = event of getting a perfect square then
P(B)= 1/3
Let A = {4, 5, 6} (i.e. it is known that the toss resulted in a number greater than 3). A is a
subset of S. To get the probability of B given that A has occurred, new weights are assigned to the
elements of A proportional to their original weights such that their sum is 1. Assigning a weight
of w to the odd number in A and a weight of 2w to the 2 even numbers, we have 5w = 1 or w =
1/5. Relative to A, B contains the single element 4, i.e. B/A = {4}. Thus,
P(A) = 2/5.

2
 This example illustrates that events may have different probabilities when considered
relative to different sample spaces.
Now,
2 2 9 P( A  B)
P( B / A) = = = ,
5 59 P( A)
where P ( A  B ) and P(A) are found from the original sample space S.

Home work:
Review the theorems and definitions regarding conditional probability.

1.5 BAYES’ Rule

Let {B1, B2, … Bn} be a set of events forming a partition of the sample space S, where
P(Bi)≠0, for i = 1, 2, …, n. Let A be any event of S such that P(A) ≠ 0. Then, for k = 1, 2, …n,

P( Bk  A) P( Bk ) P( A / Bk )
P( Bk / A) = n
= n

 P( B
i =1
i  A)  P( B ) P( A / B )
i =1
i i

Example 3: Three members of a county club have been nominated for the office of president. The
probability that Engr Adamu will be elected is 0.3, the probability that Hajiya Ramat will be elected
is 0.5, and the probability that Dr. Femi will be elected is 0.2 Should Engr Adamu be elected, the
probability for an increase in membership fees is 0.8. Should Hajiya Ramat or Dr. Femi be elected,
the corresponding probabilities for an increase in fees are 0.1 and 0.4.if someone is considering
joining the club but delays his decision for several weeks only to find out that the fees have been
increased, what is the probability that Dr. Femi was elected president of the club?.

Solution : Consider the following events:

A: the person elected increased fees,
B1: Engr Adamu is elected,
B2: Hajiya Ramat is elected,
B3: Dr. Femiis elected.

Using Bayes’ rule, we write

p ( B3  A)
p ( B3 / A) =
P ( B1  A) + P ( B2  A) + P ( B3  A)

Now,
P( B1  A) = P( B1 ) P( A / B1 ) = (0.3)(0.8) = 0.24
P( B2  A) = P( B2 ) P( A / B2 ) = (0.5)(0.1) = 0.05
P( B3  A) = P( B3 ) P( A / B3 ) = (0.2)(0.4) = 0.08

3
Hence,
0.08 8
P( B3 / A) = = = 0.2
0.24 + 0.05 + 0.08 37
In view of the fact that fees have increased, this result suggests that Dr. Femi is probably not the
President of the club.

2.0 RANDOM VARIABLES

The term statistical experiment has been used to describe any process by which several
chance measurements are obtained. Often, however, we are not interested in the details associated
with each sample point but only in some numerical description of the outcome. For example, the
sample space giving a detailed description of each possible outcome when a coin is tossed three
times is given by:

S = HHH , HHT , HTH , THH , HTT , HTT ,THT , TTH , TTT 

If one is concerned only with the number of heads that fall, then a numerical value of 0, 1, 2, or 3
will be assigned to each sample point.
The numbers 0, 1, 2, and 3 are random observations determined by the outcome of the
experiment. They may be looked upon as the values assumed by some random variables X, which
in this case represents the number of heads when a coin is tossed three times. Any one of the values
of X is denoted by the corresponding small letter, x in this case.

2.1 Discrete Probability distributions

A discrete random variable assumes each of its values with a certain probability. In the
case of tossing a coin three times, the variable X, representing the number of heads, assumes the
value 2 with probability 3/8, since three of the eight equally likely sample points result in two
heads and one tail.
Frequently, it is convenient to represent all the probabilities of a random variable X by a
formula. Such a formula would necessarily be a function of the numerical values x that is denoted
by f(x), g(x), r(x), etc. Hence, we can write:
F(x) = P(X=x), e.g. f(3) = P(X=3).
Hence, the function f(x) is a probability function or a probability distribution of the discrete
random variable X if, for each possible x,
1. f (x )  0 ,
2.  f ( x) = 1 ,
x

3. P( X = x) = f ( x) .

Example 1: Find the probability distribution of the sum of the numbers when a pair of dice is
tossed.
Solution: Let X be a random variable whose values x are the possible totals. Then x can be any
integer from 2 to 12. Two dice can fall in (6)(6) = 36 ways, each with probability 1/36. Then

4
P(X=3) = 2/36, since a total of 3 can occur in only tow ways (i.e. 1 2 and 2 1). Consideration of
other cases leads to the following probability distribution:

X 2 3 4 5 6 7 8 9 10 11 12
f(x) 1/36 2/36 3/36 4/36 5/36 6/36 5/36 4/36 3/36 2/36 1/36

Example 2: Find a formula for the probability distribution of the number of heads when a coin is
tossed four times.
Solution: Since there are 24 = 16 points in the sample space representing equally likely
outcomes, the denominator for all probabilities will be 16. To obtain the number of ways of getting,
say 3 heads, we need to consider the number of ways of partitioning four outcomes into two cells
 4
with three heads assigned to one cell and a tail into the other. This can be done in   = 4 ways.
 
3
4
In general, x heads and 4 – x tails can occur in   ways, where x can be 0, 1, 2, 3, or 4. Thus, the
 
x
probability distribution f(x) = P(X=x) is
 4
 
f (x ) =   ,
x
x = 0, 1, 2, 3, 4.
16
The cumulative distribution F(x) of a discrete random variable X with probability
distribution f(x) is given by
F ( x) = P( X  x) =  f (t ) .
tx

Example 3: Find the cumulative distribution of the random variable X in example 2 above. Using
F(x), verify that f(2) = 3/8.
Solution: Direct calculations of the probability distribution of example 2 gives f(0) = 1/16;
f(1) = 1/4; f(2) = 3/8; f(3) = 1/4; and f(4) = 1/6. Hence,
F(0) = f(0) = 1/16, _______________________ (1)
F(1) = f(0) + f(1) = 5/16, __________________ (2)
F(2) = f(0) + f(1) + f(2) = 11/16, ____________ (3)
F(3) = f(0) + f(1) + f(2) + f(3) = 15/16, _______ (4)
F(4) = f(0) + f(1) + f(2) + f(3) + f(4) = 1 ______ (5)
Therefore,

5
 0 for x  0
1
 for 0  x  1
16
5
 for 1  x  2
16
F ( x) = 
11 for 2  x  3
16
15
 for 3  x  4
16
1 for x  4
Now, f(2) = F(2) – F(1) = (11/16) – (5/16) = (3/8). (i.e. from eqn (3) and (2).

2.2 Continuous Probability Distributions:

A continuous random variable has a probability of zero assuming exactly any of its values.
Consequently, its probability distribution cannot be given in tabular form. However, it does have
a formula.
In dealing with continuous probability distribution of X, f(x) is usually called the density
function. Since X is defined over a continuous sample space, it is possible for f(x) to have a finite
number of discontinuities. However, most density functions that have practical applications are
continuous and their graph may take any one of the forms in discussed later in this course.
A probability density function is constructed so that the area under its curve bounded by
the x-axis is equal to 1 when computed over the range of x for which f(x) is defined.
Definitions
1. The function f(x) is probability density function for the continuous random variable X, defined
over the set of real numbers R, if:
i. f ( x)  0 for all x R

ii. 
−
f ( x)dx = 1
P(a  x  b) = a f ( x)dx
b
iii.

Example 1: Let the random variable x have the probability density function:
x2
f ( x) = −1  x  2
3
= 0 elsewhere
a) verify condition (ii) of definition above
b) find P(0  x  1)

Solution:
 x2  x3  2 8 1
 f ( x)dx = −1 dx =   = + = 1
2
1.
 9 −1 9 9
−
3

6
 x2  x3  1 1
P(0  X  1) = 0 dx =   = .
1
2.
3  9 0 9
2. The cumulative distribution F(x) of a continuous random variable X with density function f(x)
is given by:

F ( x) = P( X  x) = − f (t )dt.

- as an immediate consequence of this definition, one can write:

P(a  x  b) = F (b) − F (a) , and
dF ( x)
P( x) = , if the derivative exists.
dx
Example 2: For the density function of example 1 above, find F(x) and use it to evaluate P(0<X≤1).

Solution:
 t2
x  t 3  x x3 + 1
F ( x) =  f (t )dt =  dt =   =
− −1 3
 9  − 1 9
Therefore,
2 1 1
P(0  X  1) = F (1) − F (0) = − =
9 9 9
- which agrees with the result obtained by using the density function in example 1 above.

3.0 SPECIFIC DISCRETE PROBABILITY DISTRIBUTIONS

3.1 Uniform Distribution

If the RV, X assumes the values x1, x2, . . ., xk, with equal probability, then it is a uniform
distribution, given by:
f(x; k) = 1/k, x = x1, x2, . . ., xk.
The mean and variance of the discrete uniform distribution f(x; k) are:
k k

 xi  (x − )
2
i
= i =1
, and  2 = i =1

k k

Ex. 1 Find a formula or the distribution o the RV, X representing the number on a tag drawn at
random from a box containing 10 tags numbered 1 to 10. What is the probability that the number
drawn is less than 4? Sketch the histogram for this probability distribution (pd) and find the mean
and variance

7
Solution: Since the RV, X assumes its values with equal probability, the distribution is a uniform
one.
Now, S = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10}, and since each element has a probability of 1/10,
f(x; 10) = 1/10, x = 1, 2, . . ., 10.
Also, P(X = 4) = 3 × (1/10) = 3/10.
f(x; 10)

1/10

1 2 3 4 5 6 7 8 9 10 x

Fig. 3.1: Typical histogram for discrete uniform probability distribution.

1 + 2 + ..... + 10
= = 5.5
10
(1 − 5.5) 2 + (2 − 5.5) 2 + ..... + (10 − 5.5) 2
2 = = 8.25
10

3.2 Binomial Distribution

If a binomial trial results in a success with probability p and a failure with a probability q
= 1 – p, then the pd of the binomial RV, X, for the number of success in n independent trials is:
 n
b( x; n, p) =   p x q n − x , x = 0, 1, 2, . . . ., n.
 x
The mean and variance o the binomial distribution b(x; n, p) are: μ = np and σ 2 = npq.

Ex. 2 The probability that a certain kind of component will survive a given shock test is 3/4. Find
the probability that exactly two of the next four components tested survived.
Solution: Assuming the tests are independent and p = 3/4 for each of the four tests, then:
2 2
 4  3   1  4! 32 27
b(2;4, 3 4) =      = . 4 = .
 
2  4   4  2!2! 4 128

Ex. 3 The probability that a certain component survives an impact test is 0.4. If 15 components
tested shattered under the test, what is the probability that (i) at least 10 components survived, (ii)
from 3 to 8 survived, and (iii) exactly 5 components survived?
Using Chebyshev’s theorem, find and interpret the interval μ ± 2σ.

Solution: Let X be the number o components that survived the test, then:
(i) P(X ≥ 10) = 1 – P(X < 10)
9 9
= 1 −  b( x;15,0.4) (from tables,  b( x;15,0.4) is found to be 0.9662).
x =0 x =0

8
 = 1 – 0.9662 = 0.0338.

8
(ii) P(3  X  8) =  b( x;15,0.4)
x =3
8 2
=  b( x;15,0.4) −  b( x;15,0.4) = 0.9050 − 0.0271, (from tables)
x =0 x =0
= 0.8779.

(iii) P(X = 5) = b(5; 15, 0.4)

5 4
=  b( x;15,0.4) −  b( x;15,0.4) = 0.4032 − 0.2173 = 0.1859 .
x =0 x =0
Now, with n = 15, p = 0.4 and q = 0.6, we have: μ = (15)(0.4) = 6; and σ 2 = (15)(0.4)(0.6) = 3.6,
and σ = 1.897, hence:
μ ± 2σ = 6 ± (2)(1.897) or from 2.206 to 9.794.
Interpretation: Chebyshev’s theorem states that survival rate of 15 components subjected to the
impact test has a probability of at least 3/4 of falling between 2.206 and 9.794. In other words, the
probability that between 2 and 10 components out of the 15 subjected to the impact test is at least
3/4.

3.3 Poison Distribution

The pd of the Poisson RV, X, representing the number of successes occurring in a given
time interval or specified region is:
e−   x
p( x;  ) = , x = 0, 1, 2, . . . . ,
x!
where μ is the average number of successes occurring in the given time interval or specified region
end e = 2.71828…Statistical tables contain Poisson probability sums:
r
p(r;  ) =  p( x;  ) , for selected values of μ.
x =0
The mean and variance of Poisson distribution p(x; μ) both have the value μ.
Let X be a binomial RV with pd b(s; n, p), when n → ∞, p → 0, and μ = np remains
constant, then:
b(x; n, p) → p(x; μ).

Ex. 4 The average number of radioactive particles passing through a counter during 1 millisecond
in a laboratory experiment is four. What is the probability that six particles enter the counter in a
given millisecond?

Solution: Using Poisson distribution, with x = 6 and μ = 4, we find from appropriate statistical
tables that:
e −4 4 6 6 5
p (6; 4) = =  p ( x; 4) −  p ( x; 4) = 0.8893 − 0.7851 = 0.1042
6! x =0 x =0

9
Ex. 5 The average number of oil tankers arriving each day at the Kano depot is known to be 10.
The facilities at the depot can handle at most 15 tankers daily. What is the probability that on a
given day tankers will have to be sent away?

Solution: Let X be the number of tankers arriving each day, then, from tables,
15
P( X  15) = 1 − P( X  15) = 1 −  p( x; 10 ) = 1 − 0.9513 = 0.0487
x =0

3.4 Geometric distribution

If repeated independent trials can result in a success with probability p and a failure
probability q = 1 – p, then the pd of the RV, X, the number of the trial on which the first success
occurs, is given by:
g(x; p) = p qx-1, x = 1, 2, 3, . . .

Ex. 6 In a certain manufacturing process, it is known that, on the average, 1 in every 100 items is
defective. What is the probability that 5 items are inspected before a defective item is found?

Solution: Using the geometric distribution with x = 5, and p = 0.01,

g(5; 0.01) = (0.001) (0.99)4 = 0.0096.

Assignment: Other discrete distributions include: Multinomial, Negative binomial,

Hypergeometric distributions, etc, etc. Read about them.

10