On the basis of Applied Statistics and Probability for Engineers by D. C.

Montgomery - 1

Chapter 4
4-1 CONTINUOUS RANDOM VARIABLES
4-2 PROBABILITY DISTRIBUTIONS AND PROBABILITY DENSITY FUNCTIONS
4-3 CUMULATIVE DISTRIBUTION FUNCTIONS
4-4 MEAN AND VARIANCE OF A CONTINUOUS RANDOM VARIABLE
4-5 CONTINUOUS UNIFORM DISTRIBUTION
4-6 NORMAL DISTRIBUTION
4-7 NORMAL APPROXIMATION TO THE BINOMIAL AND POISSON
DISTRIBUTIONS
4-9 EXPONENTIAL DISTRIBUTION
4-10 ERLANG AND GAMMA DISTRIBUTIONS
4-11 WEIBULL DISTRIBUTION
4-12 LOGNORMAL DISTRIBUTION
Selected exercises
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 2

Discrete random variables Continuous random variables

Cumulative distribution function P(X  x ) = F( x )

P(a  X  b) = F(b) − F(a )
F
1 1 F
F(b)
F(a )
x1 x2
a b

Probability distribution Probability density function

x b
xi x1 x2 … P ( X  x ) =  f P (a  X  b ) =  f
− a
pi p1 p2 …

Probability mass function f

p i = P(X = x i ) = f ( x i )
a b
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 3

Discrete random variables Continuous random variables

Probability distribution Probability density function

xi x1 x2 … f
pi p1 p2 …

Expected value (mean)

E(X) =  x i  p i

E(X) =  x  f ( x ) dx
i −

Variance

V(X) =  (x i − E(X) ) p i =

2
V(X) = 
−
(x − E(X) )  f ( x ) dx =
2

=  x  p i − E(X)

=  x 2  f ( x ) dx − E (X) 2
2 2
i
i −
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 4

Important discrete models Important continuous models

Bernoulli trial
Binomial distribution Normal distribution
Parameters: 𝑛, 𝑝 Parameters: 𝜇, 𝜎 > 0
𝑛 1 −(𝑥−𝜇)2
𝑃(𝑋 = 𝑘) = ( ) 𝑝𝑘 (1 − 𝑝)𝑛−𝑘 , 𝑘 = 0,1, … , 𝑛
𝑘 𝑓 (𝑥 ) = ∙ 𝑒 2𝜎2 , 𝑥 ∈ ℝ
𝐸 (𝑋) = 𝑛𝑝; 𝑉 (𝑋) = 𝑛𝑝(1 − 𝑝) √2𝜋𝜎
𝐸 (𝑋) = 𝜇; 𝑉(𝑋) = 𝜎 2
Poisson process  𝜆
Poisson distribution Exponential distribution
Parameter: 𝜆 > 0 Parameter: 𝜆
𝜆𝑘 −𝜆 𝑓 (𝑥 ) = 𝜆 ∙ 𝑒−𝜆𝑥 , 𝑥 > 0
𝑃(𝑋 = 𝑘) = ∙ 𝑒 , 𝑘 = 0,1,2, …
𝑘! 𝐹 (𝑥 ) = 1 − 𝑒−𝜆𝑥 , 𝑥 > 0
𝐸 (𝑋) = 𝜆; 𝑉(𝑋) = 𝜆 1 1
𝐸 (𝑋) = ; 𝑉 (𝑋) = 2
𝜆 𝜆
Hypergeometric distribution Weibull distribution
Parameters: 𝑛, 𝑁, 𝐾 Parameter: 𝛿 > 0, 𝛽 > 0.
𝐾 𝑁−𝐾 𝛽 𝑥 𝛽−1 −(𝑥)𝛽
( )( ) 𝑓 (𝑥 ) = ( ) ∙𝑒 𝛿 , 𝑥 >0
𝑃(𝑋 = 𝑘) = 𝑘 𝑛 − 𝑘 , 𝛿 𝛿
𝑁 𝑥 𝛽
( ) −( )
𝑛 𝐹 (𝑥 ) = 1 − 𝑒 𝛿
𝑘 = max{0, 𝑛 + 𝐾 − 𝑁} , … , min{𝐾, 𝑛}, 1 2 1
2

𝐾 𝐾 𝐾 𝑁−𝑛 𝐸(𝑋) = 𝛿Γ (1 + ) , 𝑉(𝑋) = 𝛿 2 Γ (1 + ) − 𝛿 2 (Γ (1 + ))

𝐸 (𝑋) = 𝑛 ∙ ; 𝑉 (𝑋) = 𝑛 ∙ ∙ (1 − ) ∙ ( ) 𝛽 𝛽 𝛽
𝑁 𝑁 𝑁 𝑁−1
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 5

Connection between the distributions

When 𝛽 = 1, the Weibull distribution is identical to the exponential distribution.

 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 6

4-1 CONTINUOUS RANDOM VARIABLES

Previously, we discussed the measurement of the current in a thin copper wire. We noted that the
results might differ slightly in day-to-day replications because of small variations in variables that
are not controlled in our experiment—changes in ambient temperatures, small impurities in the
chemical composition of the wire, current source drifts, and so forth.
Another example is the selection of one part from a day’s production and very accurately measuring
a dimensional length. In practice, there can be small variations in the actual measured lengths due to
many causes, such as vibrations, temperature fluctuations, operator differences, calibrations, cutting
tool wear, bearing wear, and raw material changes. Even the measurement procedure can produce
variations in the final results.
In these types of experiments, the measurement of interest—current in a copper wire
experiment, length of a machined part—can be represented by a random variable.
It is reasonable to model the range of possible values of the random variable by an
interval (finite or infinite) of real numbers.
For example, for the length of a machined part, our model enables the measurement from
the experiment to result in any value within an interval of real numbers.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 7

Because the range is any value in an interval, the model provides for any precision in
length measurements.
However, because the number of possible values of the random variable 𝑋 is uncountably
infinite, 𝑋 has a distinctly different distribution from the discrete random variables
studied previously.
The range of 𝑋 includes all values in an interval of real numbers; that is, the range of 𝑋 can
be thought of as a continuum.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 8

4-2 PROBABILITY DISTRIBUTIONS AND PROBABILITY DENSITY FUNCTIONS

Density functions are commonly used in engineering to describe physical systems.
For example, consider the density of a loading on a long, thin beam. For any point 𝑥 along the beam,
the density can be described by a function (in grams/cm).
Intervals with large loadings correspond to large values for the function.
The total loading between points 𝑎 and 𝑏 is determined as the integral of the density function from 𝑎
to 𝑏.
This integral is the area under the density function over this interval, and it can be loosely
interpreted as the sum of all the loadings over this interval.
Similarly, a probability density function 𝑓(𝑥) can be used to describe the probability
distribution of a continuous random variable 𝑋.
The probability that 𝑋 is between 𝑎 and 𝑏 is determined as the integral of 𝑓 from 𝑎 to 𝑏.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 9

Definition
For a continuous random variable 𝑋, a probability density function is a function such that
𝑏

𝑃(𝑎 ≤ 𝑋 ≤ 𝑏) = ∫ 𝑓(𝑥) 𝑑𝑥 (4 − 1)
𝑎
A probability density function is zero for 𝑥 values that cannot occur and it is assumed to
be zero wherever it is not specifically defined.
Important properties of the probability density function
𝑓(𝑥) ≥ 0
∞

∫ 𝑓(𝑥) 𝑑𝑥 = 1
−∞
Define the probability density function of a continuous random variable 𝑋
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 10

For a continuous random variable 𝑋 and any value 𝑥

𝑃(𝑋 = 𝑥 ) = 0.
In practice, when a particular current measurement is observed, such as 14.47
milliamperes, this result can be interpreted as the rounded value of a current
measurement that is actually in a range such as 14.465 ≤ 𝑥 ≤ 14.475.
Therefore, the probability that the rounded value 14.47 is observed as the value for 𝑋 is
the probability that 𝑋 assumes a value in the interval [14.465,14.475], which is not zero.
Similarly, because each point has zero probability, one need not distinguish between
inequalities such as < or ≤ for continuous random variables.
If 𝑋 is a continuous random variable, for any 𝑥1 and 𝑥2
𝑃(𝑥1 ≤ 𝑋 ≤ 𝑥2 ) = 𝑃(𝑥1 < 𝑋 ≤ 𝑥2 ) = 𝑃(𝑥1 ≤ 𝑋 < 𝑥2 ) = 𝑃(𝑥1 < 𝑋 < 𝑥2 ) (4 − 2)
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 11

EXAMPLE 4-1
Let the continuous random variable 𝑋 denote the current measured in a thin copper
wire in milliamperes.
Assume that the range of 𝑋 is [0, 20 mA], and assume that the probability density
function of 𝑋 is

0.05 0 ≤ 𝑥 ≤ 20
𝑓(𝑥 ) = {
0 otherwise

What is the probability that a current measurement is less than 10 milliamperes?

The probability requested is indicated by the shaded area.
10

𝑃(𝑋 < 10) = ∫ 0.05 𝑑𝑥 = 0.5

0
As another example,
20

𝑃(5 < 𝑋 < 20) = ∫ 0.05 𝑑𝑥 = 0.75

5
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 12

EXAMPLE 4-2
Let the continuous random variable 𝑋 denote the diameter of a hole drilled in a sheet
metal component.
The target diameter is 12.5 millimeters.
Most random disturbances to the process result in larger diameters.
Historical data show that the distribution of 𝑋 can be modeled by a probability density
function
−20∙(𝑥−12.5)
𝑓(𝑥 ) = { 20 ∙ 𝑒 12.5 ≤ 𝑥
0 otherwise

If a part with a diameter larger than 12.60 millimeters is scrapped, what proportion of
parts is scrapped?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 13

A part is scrapped if 𝑋 > 12.60

Now,
∞
∞
𝑃(𝑋 > 12.6) = ∫ 20 ∙ 𝑒 −20∙(𝑥−12.5) 𝑑𝑥 = [−𝑒 −20∙(𝑥−12.5) ]12.6 = 0.135
12.6
What proportion of parts is between 12.5 and 12.6 millimeters?
Now,
12.6
12.6
𝑃(12.5 < 𝑋 < 12.6) = ∫ 20 ∙ 𝑒 −20∙(𝑥−12.5) 𝑑𝑥 = [−𝑒 −20∙(𝑥−12.5) ]12.5 = 0.865
12.5
Because the total area under 𝑓 equals 1, we can also calculate
𝑃(12.5 < 𝑋 < 12.6) = 1 − 𝑃(𝑋 > 12.6) = 1 − 0.135 = 0.865
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 14

4-3 CUMULATIVE DISTRIBUTION FUNCTIONS

An alternative method to describe the distribution of a discrete random variable can also
be used for continuous random variables.
Definition
The cumulative distribution function of a continuous random variable 𝑋 is
𝐹 (𝑥 ) = 𝑃(𝑋 ≤ 𝑥 ) (4 − 3)
for −∞ < 𝑥 < ∞.
Connection between the cumulative distribution function and the probability density
function:
𝑥

𝐹 (𝑥 ) = ∫ 𝑓(𝑢) 𝑑𝑢
−∞
Define the cumulative distribution function of a continuous random variable
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 15

EXAMPLE 4-3
For the copper current measurement in Example 4-1, the cumulative distribution
function of the random variable 𝑋 consists of three expressions.
0.05 0 ≤ 𝑥 ≤ 20
𝑓 (𝑥 ) = {
0 otherwise
therefore,
𝑥 𝑥

𝐹 (𝑥 ) = ∫ 𝑓 = ∫ 0 𝑑𝑢 = 0, 𝑖𝑓 𝑥<0
−∞ −∞
𝑥 𝑥

𝐹 (𝑥 ) = ∫ 𝑓 = ∫ 0.05 𝑑𝑢 = 0.05 ∙ 𝑥, 𝑖𝑓 0 ≤ 𝑥 < 20

−∞ 0
𝑥 20 𝑥

𝐹 (𝑥 ) = ∫ 𝑓 = ∫ 0.05 𝑑𝑢 + ∫ 0 𝑑𝑢 = 1, 𝑖𝑓 20 ≤ 𝑥
−∞ 0 20
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 16

Notice that a continuous random variable has a continuous cumulative distribution

function.
Thus, in the definition of 𝐹 any < can be changed to ≤ and vice versa.
For a discrete random variable, 𝐹 is not a continuous function.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 17

EXAMPLE 4-4
For the drilling operation in Example 4-2, 𝐹 consists of two expressions.
−20∙(𝑥−12.5)
𝑓 (𝑥 ) = {20 ∙ 𝑒 12.5 ≤ 𝑥
0 otherwise
therefore,
𝑥 𝑥

𝐹 (𝑥 ) = ∫ 𝑓 = ∫ 0 𝑑𝑢 = 0, 𝑖𝑓 𝑥 < 12.5
−∞ −∞
𝑥 𝑥

𝐹 (𝑥 ) = ∫ 𝑓 = ∫ 20 ∙ 𝑒 −20∙(𝑢−12.5) 𝑑𝑢 = 1 − 𝑒 −20∙(𝑥−12.5) , 𝑖𝑓 12.5 ≤ 𝑥

−∞ 12.5
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 18

The probability density function of a continuous random variable can be determined from
the cumulative distribution function by differentiating.
Recall that the fundamental theorem of calculus states that
𝑥
𝑑
∫ 𝑓(𝑢) 𝑑𝑢 = 𝑓(𝑥)
𝑑𝑥
−∞
Then, given 𝐹
𝑑
𝑓(𝑥 ) = 𝐹(𝑥)
𝑑𝑥
as long as the derivative exists.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 19

EXAMPLE 4-5
The time until a chemical reaction is complete (in milliseconds) is approximated by the
cumulative distribution function
0 𝑥<0
𝐹 (𝑥 ) = {
1 − 𝑒 −0.01∙𝑥 0 ≤ 𝑥
Determine the probability density function of 𝑋.
What proportion of reactions is complete within 200 milliseconds?
Using the result that the probability density function is the derivative of the 𝐹, we obtain
0 𝑥<0
𝑓(𝑥 ) = {
0.01 ∙ 𝑒 −0.01∙𝑥 0 ≤ 𝑥
The probability that a reaction completes within 200 milliseconds is
𝑃(𝑋 < 200) = 𝐹 (200) = 1 − 𝑒 −2 = 0.8647
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 20

4-4 MEAN AND VARIANCE OF A CONTINUOUS RANDOM VARIABLE

The mean and variance of a continuous random variable are defined similarly to a discrete
random variable. Integration replaces summation in the definitions.
Definition
Suppose 𝑋 is a continuous random variable with probability density function 𝑓(𝑥).
The mean or expected value of 𝑋, denoted as or 𝐸(𝑋), is
∞

𝜇 = 𝐸 (𝑋) = ∫ 𝑥 ∙ 𝑓(𝑥) 𝑑𝑥 (4 − 4)
−∞
The variance of 𝑋, denoted as 𝑉(𝑋) or 𝜎 is
2
∞

𝜎 2 = 𝑉 (𝑋) = ∫ (𝑥 − 𝜇)2 ∙ 𝑓(𝑥) 𝑑𝑥

−∞
The standard deviation of 𝑋 is
𝜎 = √𝜎 2
Give the expected value and the variance of a continuous random variable 𝑋
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 21

It can be proved that

∞

𝑉 (𝑋) = ∫ 𝑥 2 ∙ 𝑓(𝑥) 𝑑𝑥 − 𝜇2
−∞
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 22

The equivalence of the two formulas for variance can be derived as one, as was done for
discrete random variables.
EXAMPLE 4-6
Determine the mean of 𝑋 in the copper current measurement in Example 4-1.
0.05 0 ≤ 𝑥 ≤ 20
𝑓 (𝑥 ) = {
0 otherwise
therefore
∞ 20 20
𝑥2
𝐸 (𝑋) = ∫ 𝑥 ∙ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑥 ∙ 0.05 𝑑𝑥 = [0.05 ∙ ] = 10
2 0
−∞ 0
∞ 20 20
2 2
(𝑥 − 10)3
𝑉 (𝑋) = ∫ (𝑥 − 10) ∙ 𝑓(𝑥) 𝑑𝑥 = ∫ (𝑥 − 10) ∙ 0.05 𝑑𝑥 = [0.05 ∙ ] = 33.33
3 0
−∞ 0
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 23

The expected value of a function ℎ(𝑋) of a continuous random variable is defined similarly
to a function of a discrete random variable.
Expected Value of a Function of a Continuous Random Variable
If 𝑋 is a continuous random variable with probability density function 𝑓(𝑥),
∞

𝐸 (ℎ(𝑋)) = ∫ ℎ(𝑥) ∙ 𝑓(𝑥) 𝑑𝑥 (4 − 5)

−∞
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 24

EXAMPLE 4-7
In Example 4-1, 𝑋 is the current measured in milliamperes.
What is the expected value of the squared current?
Now, ℎ(𝑋) = 𝑋 2
Therefore,
∞ 20 20
3
𝑥
𝐸 (ℎ(𝑋)) = 𝐸 (𝑋 2 ) ∫ 𝑥 2 ∙ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑥 2 ∙ 0.05 𝑑𝑥 = [0.05 ∙ ] = 133.33
3 0
−∞ 0
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 25

EXAMPLE 4-8
For the drilling operation in Example 4-2,
−20∙(𝑥−12.5)
𝑓 (𝑥 ) = {20 ∙ 𝑒 12.5 ≤ 𝑥
0 otherwise
The mean of 𝑋 is
∞ ∞

𝐸 (𝑋) = ∫ 𝑥 ∙ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑥 ∙ 20 ∙ 𝑒 −20∙(𝑥−12.5) 𝑑𝑥

−∞ 12.5
Integration by parts can be used to show that
−20∙(𝑥−12.5) ∞
𝑒
𝐸 (𝑋) = [−𝑥 ∙ 𝑒 −20∙(𝑥−12.5) − ] = 12.55
20 12.5
Similarly,
∞ ∞

𝑉 (𝑋) = ∫ (𝑥 − 12.55)2 ∙ 𝑓(𝑥) 𝑑𝑥 = ∫ (𝑥 − 12.55)2 ∙ 20 ∙ 𝑒 −20∙(𝑥−12.5) 𝑑𝑥 = 0.0025

−∞ 12.5
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 26

4-5 CONTINUOUS UNIFORM DISTRIBUTION

The simplest continuous distribution is analogous to its discrete counterpart.
Definition
A continuous random variable 𝑋 with probability density function
1
𝑓(𝑥 ) = , 𝑎≤𝑥≤𝑏 (4 − 6)
𝑏−𝑎
is a continuous uniform random variable.

Give the probability density function of the continuous uniform random variable
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 27

The cumulative distribution function of a continuous uniform random variable is obtained

by integration.
If 𝑎 < 𝑥 < 𝑏,
𝑥
1 1
𝐹 (𝑥 ) = ∫ 𝑑𝑡 = ∙ (𝑥 − 𝑎)
𝑏−𝑎 𝑏−𝑎
𝑎
Therefore, the complete description of the cumulative distribution function of a
continuous uniform random variable is
0 𝑥<𝑎
𝑥−𝑎
𝐹 (𝑥 ) = { 𝑎≤𝑥<𝑏
𝑏−𝑎
1 𝑏≤𝑥
An example of 𝐹 for a continuous uniform random variable:
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 28

The mean of the continuous uniform random variable 𝑋 is

𝑏 𝑏
1 1 𝑥2 𝑎+𝑏
( )
𝐸 𝑋 = ∫𝑥 ∙ 𝑑𝑥 = [ ∙ ] =
𝑏−𝑎 𝑏−𝑎 2 𝑎 2
𝑎
The variance of 𝑋 is
𝑏
2 3 𝑏
𝑎+𝑏 1 1 1 𝑎+𝑏 (𝑏 − 𝑎)2
𝑉(𝑋) = ∫ (𝑥 − ) ∙ 𝑑𝑥 = [ ∙ ∙ (𝑥 − ) ] =
2 𝑏−𝑎 𝑏−𝑎 3 2 𝑎
12
𝑎
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 29

EXAMPLE 4-9
Let the continuous random variable 𝑋 denote the current measured in a thin copper
wire in milliamperes.
Assume that the range of 𝑋 is [0, 20 mA], and assume that the probability density
function of 𝑋 is
0.05 0 ≤ 𝑥 ≤ 20
𝑓 (𝑥 ) = {
0 otherwise
What is the probability that a measurement of current is between 5 and 10
milliamperes?
10

𝑃(5 < 𝑋 < 10) = ∫ 0.05 𝑑𝑥 = 0.25

5

The mean and variance formulas can be applied with 𝑎 = 0 and 𝑏 = 20.
Therefore, 𝐸 (𝑋) = 10, 𝑉 (𝑋) = 33.33, 𝜎 = 5.77𝑚𝐴
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 30

4-6 NORMAL DISTRIBUTION

Undoubtedly, the most widely used model for the distribution of a random variable is a
normal distribution.
Whenever a random experiment is replicated, the random variable that equals the
average (or total) result over the replicates tends to have a normal distribution as the
number of replicates becomes large.
For example, an automotive engineer may plan a study to average pull-off force
measurements from several connectors. If we assume that each measurement results
from a replicate of a random experiment, the normal distribution can be used to make
approximate conclusions about this average.
Sometimes the central limit theorem is less obvious. For example, assume that the
deviation (or error) in the length of a machined part is the sum of a large number of
infinitesimal effects, such as temperature and humidity drifts, vibrations, cutting angle
variations, cutting tool wear, bearing wear, rotational speed variations, mounting and
fixturing variations, variations in numerous raw material characteristics, and variation in
levels of contamination.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 31

If the component errors are independent and equally likely to be positive or negative, the
total error can be shown to have an approximate normal distribution.
Furthermore, the normal distribution arises in the study of numerous basic physical
phenomena. For example, the physicist Maxwell developed a normal distribution from
simple assumptions regarding the velocities of molecules.
The theoretical basis of a normal distribution is mentioned to justify the somewhat
complex form of the probability density function.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 32

Random variables with different means and variances can be modeled by normal
probability density functions with appropriate choices of the center and width of the
curve.
The value of 𝐸 (𝑋) = 𝜇 determines the center of the probability density function and the
value of 𝑉 (𝑋) = 𝜎 2 determines the width.
The figure illustrates several normal probability density functions with selected values of
𝜇 and 𝜎 2 .

Each has the characteristic symmetric bell-shaped curve, but the centers and dispersions
differ.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 33

The following definition provides the formula for normal probability density functions.
Definition
A random variable 𝑋 with probability density function
1 −(𝑥−𝜇)2
𝑓 (𝑥 ) = ∙ 𝑒 2𝜎2 , −∞ < 𝑥 < ∞ (4 − 8)
√2𝜋 ∙ 𝜎
is a normal random variable with parameters 𝜇 and 𝜎, where −∞ < 𝜇 < ∞ and 𝜎 > 0.

𝐸 (𝑋) = 𝜇 and 𝑉(𝑋) = 𝜎 2 (4 − 9)

and the notation 𝑁(𝜇, 𝜎 2 ) is used to denote the distribution.

Give the probability density function of a normal random variable

 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 34

EXAMPLE 4-10
Assume that the current measurements in a strip of wire follow a normal distribution
with a mean of 10 milliamperes and a variance of 4 (𝑚𝐴2).
What is the probability that a measurement exceeds 13𝑚𝐴?
Let 𝑋 denote the current in milliamperes.
The requested probability can be represented as 𝑃(𝑋 > 13).
This probability is shown as the shaded area under the normal probability density
function.
Unfortunately, there is no closed-form expression for the integral of a normal
probability density function, and probabilities based on the normal distribution are
typically found numerically or from a table.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 35

For any normal random variable,

𝑃(𝜇 − 𝜎 < 𝑋 < 𝜇 + 𝜎) = 0.6827
𝑃(𝜇 − 2𝜎 < 𝑋 < 𝜇 + 2𝜎) = 0.9545
𝑃(𝜇 − 3𝜎 < 𝑋 < 𝜇 + 3𝜎) = 0.9973
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 36

From the symmetry of 𝑓,

𝑃(𝑋 > 𝜇) = 𝑃(𝑋 < 𝜇) = 0.5
The probability density function decreases as 𝑥 moves farther from 𝜇. Consequently, the
probability that a measurement falls far from 𝜇 is small, and at some distance from 𝜇 the
probability of an interval can be approximated as zero.
The area under a normal probability density function beyond 3𝜎 from the mean is quite
small. This fact is convenient for quick, rough sketches of a normal probability density
function.
The sketches help us determine probabilities.
Because more than 0.9973 of the probability of a normal distribution is within the interval
(𝜇 − 3𝜎, 𝜇 + 3𝜎), 6𝜎 is often referred to as the width of a normal distribution.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 37

Machine and process capability (Statistical Process Control)

𝑈𝑆𝐿= upper specification limit
𝐿𝑆𝐿= lower specification limit
𝑈𝑆𝐿 − 𝐿𝑆𝐿
𝐶𝑚
6𝜎𝑠ℎ𝑜𝑟𝑡−𝑡𝑒𝑟𝑚
𝑈𝑆𝐿 − 𝜇 𝜇 − 𝐿𝑆𝐿
𝐶𝑚𝑘 𝑚𝑖𝑛 { , }
3𝜎𝑠𝑡 3𝜎𝑠𝑡
𝑈𝑆𝐿 − 𝐿𝑆𝐿
𝐶𝑝
6𝜎𝑙𝑜𝑛𝑔−𝑡𝑒𝑟𝑚 

𝑈𝑆𝐿 − 𝜇 𝜇 − 𝐿𝑆𝐿 − 6 − 3 − + + 3 + 6
𝐶𝑝𝑘 𝑚𝑖𝑛 { , }
3𝜎𝑙𝑡 3𝜎𝑙𝑡
ATH FTH

𝑚 is for “machine”
𝑝 is for “process”
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 38

Examples

𝑪𝒎 = 𝟏, 𝟔𝟕

𝑪𝒎𝒌 = 𝟏, 𝟔𝟕 

− 6 − 3 − + + 3 + 6

ATH FTH

𝑪𝒎 = 𝟏, 𝟔𝟕

𝑪𝒎𝒌 = 𝟏, 𝟑𝟑 

− 6 − 3 − + + 3 + 6

ATH FTH
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 39

𝑪𝒎 = 𝟏, 𝟔𝟕

𝑪𝒎𝒌 = 𝟏 

− 6 − 3 − + + 3 + 6

ATH FTH

𝑪𝒎 = 𝟏, 𝟑𝟑

𝑪𝒎𝒌 = 𝟏, 𝟑𝟑 

− 6 − 3 − + + 3 + 6

ATH FTH
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 40

𝑪𝒎 = 𝟏, 𝟑𝟑

𝑪𝒎𝒌 = 𝟏 

− 6 − 3 − + + 3 + 6

ATH FTH

𝑪𝒎 = 𝟏

𝑪𝒎𝒌 = 𝟏 

− 6 − 3 − + + 3 + 6

ATH FTH
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 41

Definition
A normal distribution with
𝜇 = 0 and 𝜎 = 1
is called a standard normal random variable and is denoted as 𝑍.
The probability density function of a The cumulative distribution function of a
standard normal random variable is standard normal random variable is
1 −𝑥 2 Φ(𝑧) = 𝑃(𝑍 ≤ 𝑧)
𝜑(𝑥 ) = ∙𝑒 2
√2𝜋
On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 42
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 43

https://stattrek.com/online-calculator/normal.aspx
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 44

EXAMPLE 4-11
Assume 𝑍 is a standard normal random variable.
Use the table to determine probabilities.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 45

EXAMPLE 4-12
The following calculations are shown pictorially.
In practice, a probability is often rounded to one or two significant digits.
(1) 𝑃(𝑍 > 1.26) = 1 − 𝑃(𝑍 ≤ 1.26) = 1 − 0.89616 = 0.10384

(2) 𝑃(𝑍 < −0.86) = 0.19490

 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 46

(3) 𝑃(𝑍 > −1.37) = 𝑃(𝑍 < 1.37) = 0.91465

(4) 𝑃(−1.25 < 𝑍 < 0.37) = 𝑃(𝑍 < 0.37) − 𝑃(𝑍 < −1.25) =
= 0.64431 − 0.10565 = 0.53866
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 47

The preceding examples show how to calculate probabilities for standard normal random
variables. Fortunately, all normal probability distributions are related algebraically.
If 𝑋 is a normal random variable with 𝐸 (𝑋) = 𝜇 and 𝑉(𝑋) = 𝜎 2 , the random variable
𝑋−𝜇
𝑍= (4 − 10)
𝜎
is a normal random variable with 𝐸 (𝑍) = 0 and 𝑉(𝑍) = 1. That is, 𝑍 is a standard normal
random variable.
Creating a new random variable by this transformation is referred to as standardizing.
The random variable 𝑍 represents the distance of 𝑋 from its mean in terms of standard
deviations.
It is the key step to calculate a probability for an arbitrary normal random variable:
𝑋−𝜇 𝑥−𝜇 𝑥−𝜇
𝐹 (𝑥 ) = 𝑃(𝑋 < 𝑥 ) = P ( < ) = Φ( )
𝜎 𝜎 𝜎
Example
𝜇 = 24, 𝜎 = 3
18 − 24
𝐹 (18) = Φ ( ) = Φ(−2) = 1 − Φ(2) = 1 − 0.9772 = 0.0228
3
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 48

EXAMPLE 4-13
Suppose the current measurements in a strip of wire are assumed to follow a normal
distribution with a mean of 10 milliamperes and a variance of 4𝑚𝐴2 .
What is the probability that a measurement will exceed 13𝑚𝐴?
Let 𝑋 denote the current in milliamperes.
The requested probability can be represented as 𝑃(𝑋 > 13).
𝑋−10
Let 𝑍 = .
2
The figures shows the relationship between the several values of 𝑋 and the transformed
values of 𝑍.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 49

13−10
𝑋 > 13 corresponds to 𝑍 > 1.5 = , therefore
2
𝑃(𝑋 > 13) = 𝑃(𝑍 > 1.5) = 1 − 𝑃(𝑍 ≤ 1.5) = 1 − 0.93319 = 0.06681

In the preceding example, the value 13 is transformed to 1.5 by standardizing, and 1.5 is
often referred to as the 𝑧-value associated with a probability.
The following summarizes the calculation of probabilities derived from normal random
variables.
Suppose 𝑋 is a normal random variable with mean 𝜇 and variance 𝜎 2 . Then,
𝑋−𝜇 𝑥−𝜇
𝑃(𝑋 ≤ 𝑥 ) = 𝑃 ( ≤ ) = 𝑃(𝑍 ≤ 𝑧) (4 − 11)
𝜎 𝜎
𝑥−𝜇
where 𝑍 is a standard normal random variable, and 𝑧 = is the 𝑧-value obtained by
𝜎
standardizing 𝑋.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 50

EXAMPLE 4-14
Continuing the previous example, what is the probability that a current measurement is
between 9 and 11 milliamperes?
9 − 10 11 − 10
𝑃(9 < 𝑋 < 11) = 𝑃 ( <𝑍< ) = 𝑃(−0.5 < 𝑍 < 0.5) =
2 2
= 𝑃(𝑍 < 0.5) − 𝑃(𝑍 < −0.5) = 0.69146 − 0.30854 = 0.38292
Determine the value for which the probability that a current measurement is below this
value is 0.98.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 51

We need the value of 𝑥 such that 𝑃(𝑋 < 𝑥) = 0.98.

By standardizing, this probability expression can be written as
𝑥 − 10
( )
𝑃 𝑋 < 𝑥 = 𝑃 (𝑍 < ) = 0.98
2
From the table 𝑃(𝑍 < 2.05) = 0.97982
Therefore
𝑥 − 10
= 2.05
2
𝑥 = 14.1 𝑚𝐴
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 52

EXAMPLE 4-15
Assume that in the detection of a digital signal the background noise follows a normal
distribution with a mean of 0 volt and standard deviation of 0.45 volt.
The system assumes a digital 1 has been transmitted when the voltage exceeds 0.9.
What is the probability of detecting a digital 1 when none was sent?
Let the random variable 𝑁 denote the voltage of noise.
The requested probability is
0.9
𝑃(𝑁 > 0.9) = 𝑃 (𝑍 > ) = 𝑃(𝑍 > 2) = 1 − 0.97725 = 0.02275
0.45
This probability can be described as the probability of a false detection.
Determine symmetric bounds about 0 that include 99% of all noise readings.
The question requires us to find 𝑥 such that 𝑃(−𝑥 < 𝑁 < 𝑥 ) = 0.99.
−𝑥 𝑥
0.99 = 𝑃(−𝑥 < 𝑁 < 𝑥 ) = 𝑃 ( <𝑍< )
0.45 0.45
From the table 𝑃(−2.58 < 𝑍 < 2.58) = 0.99, therefore
𝑥
= 2.58 ⟹ 𝑥 = 1.16
0.45
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 53

Suppose a digital 1 is represented as a shift in the mean of the noise distribution to 1.8
volts.
What is the probability that a digital 1 is not detected?
Let the random variable 𝑆 denote the voltage when a digital 1 is transmitted.
Then,
0.9 − 1.8
𝑃(𝑆 < 0.9) = 𝑃 (𝑍 < ) = 𝑃(𝑍 < −2) = 0.02275
0.45
This probability can be interpreted as the probability of a missed signal.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 54

EXAMPLE 4-16
The diameter of a shaft in an optical storage drive is normally distributed with mean
0.2508 inch and standard deviation 0.0005 inch.
The specifications on the shaft are 0.2500 ± 0.0015 inch.
What proportion of shafts conforms to specifications?
Let 𝑋 denote the shaft diameter in inches.

0.2485 − 0.2508 0.2515 − 0.2508

𝑃(0.2485 < 𝑋 < 0.2515) = 𝑃 ( <𝑍< )=
0.0005 0.0005
= 𝑃(−4.6 < 𝑍 < 1.4) = 𝑃(𝑍 < 1.4) − 𝑃(𝑍 < −4.6) = 0.91924 − 0.00000 = 0.91924
Most of the nonconforming shafts are too large, because the process mean is located
very near to the upper specification limit.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 55

If the process is centered so that the process mean is equal to the target value of 0.2500,
0.2485 − 0.2500 0.2515 − 0.2500
( )
𝑃 0.2485 < 𝑋 < 0.2515 = 𝑃 ( <𝑍< )=
0.0005 0.0005
= 𝑃(−3 < 𝑍 < 3) = 𝑃(𝑍 < 3) − 𝑃(𝑍 < −3) = 0.99865 − 0.00135 = 0.9973
By re-centering the process, the yield is increased to approximately 99.73%.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 56

4-7 NORMAL APPROXIMATION TO THE BINOMIAL AND POISSON DISTRIBUTIONS

We began our section on the normal distribution with the central limit theorem and the
normal distribution as an approximation to a random variable with a large number of
trials.
Consequently, it should not be a surprise to learn that the normal distribution can be used
to approximate binomial probabilities for cases in which 𝑛 is large.
The following example illustrates that for many
physical systems the binomial model is
appropriate with an extremely large value for 𝑛.
In these cases, it is difficult to calculate
probabilities by using the binomial distribution.
Fortunately, the normal approximation is most
effective in these cases.
The area of each bar equals the binomial
probability of 𝑥.
Notice that the area of bars can be approximated
by areas under the normal density function.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 57

EXAMPLE 4-17
In a digital communication channel, assume that the number of bits received in error
can be modeled by a binomial random variable, and assume that the probability that a
bit is received in error is 10−5 .
If 16 million bits are transmitted, what is the probability that more than 150 errors
occur?
Let the random variable 𝑋 denote the number of errors.
Then 𝑋 is a binomial random variable and
𝑃(𝑋 > 150) = 1 − 𝑃(𝑋 ≤ 150) =
150
16,000,000 ( −5 )𝑘 (
= 1−∑( ) ∙ 10 ∙ 1 − 10−5 )16,000,000−𝑘
𝑘
𝑘=0

Clearly, this probability is difficult to compute. Fortunately, the normal distribution can be
used to provide an excellent approximation in this example.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 58

Normal Approximation to the Binomial Distribution

If 𝑋 is a binomial random variable,
𝑋 − 𝑛𝑝
𝑍= (4 − 12)
√𝑛𝑝(1 − 𝑝)
is approximately a standard normal random variable.
The approximation is good for
𝑛𝑝 > 5 and 𝑛(1 − 𝑝) > 5

Recall that for a binomial variable 𝑋, 𝐸(𝑋) = 𝑛𝑝 and 𝑉(𝑋) = 𝑛𝑝(1 − 𝑝).
Consequently, the expression in Equation 4-12 is nothing more than the formula for
standardizing the random variable 𝑋.
Probabilities involving 𝑋 can be approximated by using a standard normal distribution.
The approximation is good when 𝑛 is large relative to 𝑝.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 59

EXAMPLE 4-18
The digital communication problem in the previous example is solved as follows.
𝜇 = 𝑛 ∙ 𝑝 = 16,000,000 ∙ 10−5 = 160
𝜎 = √𝑛 ∙ 𝑝 ∙ (1 − 𝑝) = √16,000,000 ∙ 10−5 ∙ (1 − 10−5 ) = 12.65:
150 − 160
𝑃(𝑋 > 150) = 𝑃 (𝑍 > ) = 𝑃(𝑍 > −0.79) = 𝑃(𝑍 < 0.79) = 0.785
12.65
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 60

EXAMPLE 4-19
Consider the transmission of bits in Example 4-18.
To judge how well the normal approximation works, assume only 𝑛 = 50 bits are to be
transmitted and that the probability of an error is 𝑝 = 0.1.
The exact probability that 2 or less errors occur is
50 50 50
𝑃(𝑋 ≤ 2) = ( ) ∙ 0.950 + ( ) ∙ 0.1 ∙ 0.949 + ( ) ∙ 0.12 ∙ 0.948 = 0.112
0 1 2
Based on the normal approximation
2−5
𝑃(𝑋 ≤ 2) = 𝑃 (𝑍 ≤ ) = 𝑃(𝑍 ≤ −1.42) = 0.08
2.12
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 61

If 𝑛𝑝 or 𝑛(1 − 𝑝) is small, the binomial distribution is quite skewed and the symmetric
normal distribution is not a good approximation.

However, a correction factor can be used that will further improve the approximation.
This factor is called a continuity correction.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 62

Recall that the binomial distribution is a satisfactory approximation to the

hypergeometric distribution when 𝑛, the sample size, is small relative to 𝑁, the size of the
population from which the sample is selected.
𝑛
A rule of thumb is that the binomial approximation is effective if < 0.1.
𝑁
𝐾
Recall that for a hypergeometric distribution 𝑝 is defined as 𝑝 = .
𝑁
That is, 𝑝 is interpreted as the number of successes in the population.
Therefore, the normal distribution can provide an effective approximation of
𝑛
hypergeometric probabilities when < 0.1, 𝑛𝑝 > 5 and 𝑛(1 − 𝑝) > 5.
𝑁
Conditions for approximating hypergeometric and binomial probabilities:

Recall that the Poisson distribution was developed as the limit of a binomial distribution
as the number of trials increased to infinity.
Consequently, it should not be surprising to find that the normal distribution can also be
used to approximate probabilities of a Poisson random variable.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 63

Normal Approximation to the Poisson Distribution

If 𝑋 is a Poisson random variable with 𝐸 (𝑋) = 𝜆 and 𝑉 (𝑋) = 𝜆
𝑋−𝜆
𝑍= (4 − 13)
√𝜆
is approximately a standard normal random variable.
The approximation is good for
𝜆 > 5.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 64

EXAMPLE 4-20
Assume that the number of asbestos particles in a squared meter of dust on a surface
follows a Poisson distribution with a mean of 1000.
If a squared meter of dust is analyzed, what is the probability that less than 950
particles are found?
This probability can be expressed exactly as
950
1000𝑘 −1000
𝑃(𝑋 ≤ 950) = ∑ ∙𝑒
𝑘!
𝑘=0
The computational difficulty is clear. The probability can be approximated as
950 − 1000
𝑃(𝑋 ≤ 950) = 𝑃 (𝑍 ≤ ) = 𝑃(𝑍 ≤ −1.58) = 0.057
√1000
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 65

4-9 EXPONENTIAL DISTRIBUTION

The discussion of the Poisson distribution defined a random variable to be the number of
flaws along a length of copper wire.
The distance between flaws is another random variable that is often of interest.
Let the random variable 𝑋 denote the length from any starting point on the wire until a
flaw is detected.
As you might expect, the distribution of 𝑋 can be obtained from knowledge of the
distribution of the number of flaws.
The key to the relationship is the following concept.
The distance to the first flaw exceeds 3 millimeters if and only if there are no flaws within
a length of 3 millimeters—simple, but sufficient for an analysis of the distribution of 𝑋.
In general, let the random variable 𝑌 denote the number of flaws in 𝑥 millimeters of wire.
If the mean number of flaws is 𝜆 per millimeter, 𝑁 has a Poisson distribution with mean
𝜆𝑥:
(𝜆𝑥)𝑘 −𝜆𝑥
𝑃(𝑌 = 𝑘) = ∙𝑒
𝑘!
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 66

We assume that the wire is longer than the value of 𝑥.

Now,
(𝜆𝑥 )0 −𝜆𝑥
𝑃(𝑋 > 𝑥 ) = 𝑃(𝑌 = 0) = ∙𝑒 = 𝑒 −𝜆𝑥
0!

Therefore,
𝐹 (𝑥 ) = 𝑃(𝑋 ≤ 𝑥 ) = 1 − 𝑒 −𝜆𝑥 , 𝑥 ≥ 0
is the cumulative distribution function of 𝑋.
By differentiating 𝐹(𝑥), the probability density function of 𝑋 is calculated to be
𝑓(𝑥 ) = 𝜆𝑒 −𝜆𝑥 , 𝑥 ≥ 0
The derivation of the distribution of 𝑋 depends only on the assumption that the flaws in
the wire follow a Poisson process.
Also, the starting point for measuring 𝑋 doesn’t matter because the probability of the
number of flaws in an interval of a Poisson process depends only on the length of the
interval, not on the location.
For any Poisson process, the following general result applies.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 67

Definition
The random variable 𝑋 that equals the distance between successive counts of a Poisson
process with mean 𝜆 > 0 is an exponential random variable with parameter 𝜆.
The probability density function of 𝑋 is
𝑓(𝑥 ) = 𝜆𝑒 −𝜆𝑥 , 0 ≤ 𝑥 < ∞ (4 − 14)
The exponential distribution obtains its
name from the exponential function in the
probability density function.
Plots of the exponential distribution for
selected values of 𝜆 are shown in the figure.
For any value of 𝜆, the exponential
distribution is quite skewed.

Give the probability density function of an

exponential random variable
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 68

The probability density function of 𝑋 is

𝐹 (𝑥 ) = 1 − 𝑒 −𝜆𝑥 , 0 ≤ 𝑥 < ∞

The following results are easily obtained and are left as an exercise.
If the random variable 𝑋 has an exponential distribution with parameter 𝜆,
1
𝜇 = 𝐸 (𝑋) =
𝜆
and
2
1
𝜎 = 𝑉(𝑋) = 2 (4 − 15)
𝜆

It is important to use consistent units in the calculation of probabilities, means, and

variances involving exponential random variables. The following example illustrates unit
conversions.
Give the cumulative distribution function of an exponential random variable
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 69

EXAMPLE 4-21
In a large corporate computer network, user log-ons to the system can be modeled as a
Poisson process with a mean of 25 log-ons per hour.
What is the probability that there are no log-ons in an interval of 6 minutes?
Let 𝑋 denote the time in hours from the start of the interval until the first log-on.
Then, 𝑋 has an exponential distribution with 𝜆 = 25 log-ons per hour.
We are interested in the probability that 𝑋 exceeds 6 minutes.
Because 𝜆 is given in log-ons per hour, we express all time units in hours.
That is, 6 minutes = 0.1 hour.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 70

𝑃(𝑋 > 0.1) = ∫ 25 ∙ 𝑒 −25∙𝑥 𝑑𝑥 = 𝑒 −25∙0.1 = 0.082

0.1
Also, the cumulative distribution function can be used to obtain the same result as
follows:
𝑃(𝑋 > 0.1) = 1 − 𝐹 (0.1) = 1 − (1 − 𝑒 −25∙0.1 ) = 𝑒 −2.5 = 0.082
What is the probability that the time until the next log-on is between 2 and 3 minutes?
Upon converting all units to hours,
0.05

𝑃(0.033 < 𝑋 < 0.05) = ∫ 25 ∙ 𝑒 −25∙𝑥 𝑑𝑥 = [−𝑒 −25∙𝑥 ]0.05

0.033 = 0.152
0.033
An alternative solution is
𝑃(0.033 < 𝑋 < 0.05) = 𝐹 (0.05) − 𝐹 (0.033) = 0.152
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 71

Determine the interval of time such that the probability that no log-on occurs in the
interval is 0.90.
The question asks for the length of time 𝑥 such that 𝑃(𝑋 > 𝑥 ) = 0.90.
Now,
0.90 = 𝑃(𝑋 > 𝑥 ) = 𝑒 −25∙𝑥
𝑥 = 0.00421 hour = 0.25 min
Furthermore, the mean time until the next log-on is
1
𝜇= = 0.04 hour = 2.4 min
25
The standard deviation of the time until the next log-on is
1
𝜎= = 0.04 hour = 2.4 min
25
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 72

In the previous example, the probability that there are no log-ons in a 6-minute interval is
0.082 regardless of the starting time of the interval.
A Poisson process assumes that events occur uniformly throughout the interval of
observation; that is, there is no clustering of events.
If the log-ons are well modeled by a Poisson process, the probability that the first log-on
after noon occurs after 12:06 pm. is the same as the probability that the first log-on after
3:00 pm. occurs after 3:06 pm.
And if someone logs on at 2:22 pm., the probability the next log-on occurs after 2:28 pm. is
still 0.082.
Our starting point for observing the system does not matter.
However, if there are high-use periods during the day, such as right after 8:00 am,
followed by a period of low use, a Poisson process is not an appropriate model for log-ons
and the distribution is not appropriate for computing probabilities.
It might be reasonable to model each of the high- and low-use periods by a separate
Poisson process, employing a larger value for 𝜆 during the high-use periods and a smaller
value otherwise.
Then, an exponential distribution with the corresponding value of 𝜆 can be used to
calculate log-on probabilities for the high- and low-use periods.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 73

Lack of Memory Property

An even more interesting property of an exponential random variable is concerned with
conditional probabilities.

EXAMPLE 4-22
Let 𝑋 denote the time between detections of a particle with a geiger counter and assume
that 𝑋 has an exponential distribution with the mean (expected value) 𝜆 = 1.4 minutes.
The probability that we detect a particle within 30 seconds of starting the counter is
1
− ∙0.5
𝑃(𝑋 < 0.5) = 𝐹 (0.5) = 1 − 𝑒 1.4
= 0.3
In this calculation, all units are converted to minutes.
Now, suppose we turn on the Geiger counter and wait 3 minutes without detecting a
particle.
What is the probability that a particle is detected in the next 30 seconds?
Because we have already been waiting for 3 minutes, we feel that we are “due.’’
That is, the probability of a detection in the next 30 seconds should be greater than 0.3.
However, for an exponential distribution, this is not true.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 74

The requested probability can be expressed as the conditional probability that

𝑃(𝑋 < 3.5|𝑋 > 3)
From the definition of conditional probability,
𝑃(3 < 𝑋 < 3.5) 𝐹 (3.5) − 𝐹(3)
𝑃(𝑋 < 3.5|𝑋 > 3) = = = 0.3
𝑃(3 < 𝑋) 1 − 𝐹(3)
After waiting for 3 minutes without a detection, the probability of a detection in the next
30 seconds is the same as the probability of a detection in the 30 seconds immediately
after starting the counter.
The fact that you have waited 3 minutes without a detection does not change the
probability of a detection in the next 30 seconds.

Example 4-22 illustrates the lack of memory property of an exponential random variable
and a general statement of the property follows.
In fact, the exponential distribution is the only continuous distribution with this property.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 75

For an exponential random variable 𝑋,

𝑃(𝑋 < 𝑡1 + 𝑡2 |𝑋 > 𝑡1 ) = 𝑃(𝑋 < 𝑡2 ) (4 − 16)

The area of region 𝐴 divided by the total area under the probability density function
𝐴 + 𝐵 + 𝐶 + 𝐷 = 1 equals 𝑃(𝑋 < 𝑡2 ).
The area of region 𝐶 divided by the area 𝐶 + 𝐷 equals 𝑃(𝑋 < 𝑡1 + 𝑡2 )|𝑋 > 𝑡1 ).
The lack of memory property implies that the proportion of the total area that is in 𝐴
equals the proportion of the area in 𝐶 and 𝐷 that is in 𝐶.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 76

The mathematical verification of the lack of memory property is left as a mind-expanding

exercise.
The lack of memory property is not that surprising when you consider the development of
a Poisson process.
In that development, we assumed that an interval could be partitioned into small intervals
that were independent.
These subintervals are similar to independent Bernoulli trials that comprise a binomial
process; knowledge of previous results does not affect the probabilities of events in future
subintervals.
An exponential random variable is the continuous analog of a geometric random variable,
and they share a similar lack of memory property.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 77

The exponential distribution is often used in reliability studies as the model for the time
until failure of a device.
For example, the lifetime of a semiconductor chip might be modeled as an exponential random
variable with a mean of 40,000 hours.
The lack of memory property of the exponential distribution implies that the device does not wear
out.
That is, regardless of how long the device has been operating, the probability of a failure in the next
1000 hours is the same as the probability of a failure in the first 1000 hours of operation.
The lifetime 𝐿 of a device with failures caused by random shocks might be appropriately modeled as
an exponential random variable.
However, the lifetime 𝐿 of a device that suffers slow mechanical wear, such as bearing wear, is
better modeled by a distribution such that 𝑃(𝐿 < 𝑡 + Δ𝑡|𝐿 > 𝑡) increases with 𝑡.
Distributions such as the Weibull distribution are often used, in practice, to model the failure time of
this type of device.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 78

4-10 ERLANG AND GAMMA DISTRIBUTIONS

4-10.1 Erlang Distribution
An exponential random variable describes the length until the first count is obtained in a
Poisson process.
A generalization of the exponential distribution is the length until 𝑟 counts occur in a
Poisson process.
The random variable that equals the interval length until 𝑟 counts occur in a Poisson
process has an Erlang random variable.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 79

EXAMPLE 4-23
The failures of the central processor units of large computer systems are often modeled
as a Poisson process.
Typically, failures are not caused by components wearing out, but by more random
failures of the large number of semiconductor circuits in the units.
Assume that the units that fail are immediately repaired and assume that the mean
number of failures per hour is 0.0001.
Let 𝑋 denote the time until four failures occur in a system.
Determine the probability that 𝑋 exceeds 40,000 hours.
Let the random variable 𝑁 denote the number of failures in 40,000 hours of operation.
The time until four failures occur exceeds 40,000 hours if and only if the number of
failures in 40,000 hours is three or less.
Therefore,
𝑃(𝑋 > 40,000) = 𝑃(𝑁 ≤ 3)
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 80

The assumption that the failures follow a Poisson process implies that N has a Poisson
distribution with
𝐸 (𝑁) = 40,000 ∙ 0.0001 = 4 failures per 40,000 hours
Therefore,
3
4𝑘 −4
𝑃(𝑋 > 40,000) = 𝑃(𝑁 ≤ 3) = ∑ ∙ 𝑒 = 0.433
𝑘!
𝑘=0

The cumulative distribution function of a general Erlang random variable 𝑋 can be

obtained from 𝑃(𝑋 ≤ 𝑥 ) = 1 − 𝑃(𝑋 > 𝑥 ) and 𝑃(𝑋 > 𝑥 ) can be determined as in the
previous example.
Then, the probability density function of 𝑋 can be obtained by differentiating the
cumulative distribution function and using a great deal of algebraic simplification.
In general, we can obtain the following result.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 81

Definition
The random variable 𝑋 that equals the interval length until r counts occur in a Poisson
process with mean 𝜆 > 0 has an Erlang random variable with parameters 𝜆 and 𝑟.
The probability density function of 𝑋 is
𝜆𝑟 𝑥 𝑟−1 𝑒 −𝜆𝑥
𝑓 (𝑥 ) = , 𝑥 > 0, 𝑟 = 1,2, … (4 − 17)
(𝑟 − 1)!
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 82

Clearly, an Erlang random variable with 𝑟 = 1 is an exponential random variable.

Probabilities involving Erlang random variables are often determined by computing a
summation of Poisson random variables as in Example 4-23.
The probability density function of an Erlang random variable can be used to determine
probabilities; however, integrating by parts is often necessary.
As was the case for the exponential distribution, one must be careful to define the random
variable and the parameter in consistent units.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 83

EXAMPLE 4-24
An alternative approach to computing the probability requested in Example 4-23 is to
integrate the probability density function of 𝑋.
That is,
∞
𝜆𝑟 𝑥 𝑟−1 𝑒 −𝜆𝑥
𝑃(𝑋 > 40,000) = ∫ 𝑑𝑥
(𝑟 − 1)!
40,000
where 𝑟 = 4 and 𝜆 = 0.0001.
Integration by parts can be used to verify the result obtained previously.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 84

An Erlang random variable can be thought of as the continuous analog of a negative

binomial random variable.
A negative binomial random variable can be expressed as the sum of 𝑟 geometric random
variables.
Similarly, an Erlang random variable can be represented as the sum of 𝑟 exponential
random variables.
Using this conclusion, we can obtain the following plausible result.
If 𝑋 is an Erlang random variable with parameters and 𝑟,
𝑟
𝜇 = 𝐸 (𝑋) =
𝜆
and
𝑟
𝜎 2 = 𝑉(𝑋) = 2 (4 − 18)
𝜆
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 85

4-10.2 Gamma Distribution

The Erlang distribution is a special case of the gamma distribution.
If the parameter 𝑟 of an Erlang random variable is not an integer, but 𝑟 > 0, the random
variable has a gamma distribution.
However, in the Erlang density function, the parameter 𝑟 appears as 𝑟 factorial.
Therefore, to define a gamma random variable, we require a generalization of the factorial
function.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 86

Definition
The gamma function is
∞

Γ(𝑟) = ∫ 𝑥 𝑟−1 𝑒 −𝑥 𝑑𝑥, 𝑟 > 0 (4 − 19)

0
It can be shown that the integral in the definition of Γ(𝑟) is finite.
Furthermore, by using integration by parts it can be shown that
Γ(𝑟) = (𝑟 − 1)Γ(𝑟 − 1).
If 𝑟 is a positive integer then
1 1 ∙ 3 ∙ … ∙ (2𝑟 − 1) 1 (−1)𝑟 ∙ 2𝑟
Γ(𝑟) = (𝑟 − 1)! Γ (𝑟 + ) = ∙ √𝜋 Γ (−𝑟 + ) = ∙ √𝜋
2 2𝑟 2 1 ∙ 3 ∙ … ∙ (2𝑟 − 1)

Some values of the gamma function:

Γ(1) = 0! = 1 1 5 1 3
Γ ( ) = √𝜋 Γ ( ) = Γ (2 + ) = √𝜋
Γ(2) = 1! = 1 2 2 2 4
3 1 1 7 1 1∙3∙5 15
Γ(3) = 2! = 2 Γ ( ) = Γ (1 + ) = √𝜋 Γ ( ) = Γ (3 + ) = ∙ √ 𝜋 = √𝜋
2 2 2 2 2 23 8
Γ(4) = 3! = 6
On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 87

Further values can be calculated as follows

Γ(𝑟) = (𝑟 − 1) ∙ Γ(𝑟 − 1)
Γ(𝑟 + 1)
Γ(𝑟) =
𝑟
Examples
Γ(1.7) 0.90864
( )
Γ 0.7 = = = 1.2981
0.7 0.7
Γ(3.5) = 2.5 ∙ Γ(2.5) = 2.5 ∙ 1.5 ∙ Γ(2.5) =
= 2.5 ∙ 1.5 ∙ 0.88623 = 3.32336
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 88

Definition
The random variable 𝑋 with probability density function
𝜆𝑟 𝑥 𝑟−1 𝑒 −𝜆𝑥
𝑓 (𝑥 ) = , 𝑥>0 (4 − 20)
Γ(𝑟)
has a gamma random variable with parameters 𝜆 > 0 and 𝑟 > 0.
If 𝑟 is an integer, 𝑋 has an Erlang distribution.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 89

It can be shown that 𝑓 satisfies the properties of a probability density function, and the
following result can be obtained.
Repeated integration by parts can be used, but the details are lengthy.
If X is a gamma random variable with parameters 𝜆 and 𝑟,
𝑟
𝜇 = 𝐸 (𝑋) =
𝜆
and
2
𝑟
𝜎 = 𝑉(𝑋) = 2 (4 − 21)
𝜆

Although the gamma distribution is not frequently used as a model for a physical system,
the special case of the Erlang distribution is very useful for modeling random
experiments.
Furthermore, the chi-squared distribution is a special case of the gamma distribution in
1 1 3
which 𝜆 = and 𝑟 equals one of the values , 1, , 2, ….
2 2 2
This distribution is used extensively in interval estimation and tests of hypotheses.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 90

4-11 WEIBULL DISTRIBUTION

As mentioned previously, the Weibull distribution is often used to model the time until
failure of many different physical systems.
The parameters in the distribution provide a great deal of flexibility to model systems in
which the number of failures increases with time (bearing wear), decreases with time
(some semiconductors), or remains constant with time (failures caused by external
shocks to the system).
Definition
The random variable 𝑋 with probability density function
𝛽 𝑥 𝛽−1 −(𝑥)𝛽
𝑓(𝑥 ) = ∙ ( ) ∙𝑒 𝛿 , 𝑥 >0 (4 − 22)
𝛿 𝛿
is a Weibull random variable with scale parameter 𝛿 > 0 and shape parameter 𝛽 > 0.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 91

The flexibility of the Weibull distribution is illustrated by the graphs of selected

probability density functions in the figure

By inspecting the probability density function, it is seen that when 𝛽 = 1, the Weibull
distribution is identical to the exponential distribution.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 92

If 𝑋 has a Weibull distribution with parameters 𝛿 and 𝛽, then the cumulative distribution
function of 𝑋 is
𝑥 𝛽
−( )
𝐹 (𝑥 ) = 1 − 𝑒 𝛿 (4 − 23)

If 𝑋 has a Weibull distribution with parameters 𝛿 and 𝛽,

1
𝜇 = 𝐸 (𝑋) = 𝛿 ∙ Γ (1 + )
𝛽
and
2
2 2
2 2
1
𝜎 = 𝑉 (𝑋) = 𝛿 ∙ Γ (1 + ) − 𝛿 ∙ (Γ (1 + )) (4 − 24)
𝛽 𝛽
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 93

EXAMPLE 4-25
The time to failure (in hours) of a bearing in a mechanical shaft is satisfactorily modeled
1
as a Weibull random variable with 𝛽 = and 𝛿 = 5000 hours.
2
Determine the mean time until failure.
From the expression for the mean,
1
𝐸 (𝑋) = 5000 ∙ Γ (1 + ) = 5000 ∙ Γ(3) = 5000 ∙ 2! = 10,000 hours
0.5
Determine the probability that a bearing lasts at least 6000 hours.
Now
1
6,000 2
−( )
𝑃(𝑋 > 6,000) = 1 − 𝐹 (6,000) = 𝑒 5,000 = 𝑒 −1.095 = 0.334
Consequently, only 33.4% of all bearings last at least 6000 hours.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 94

4-12 LOGNORMAL DISTRIBUTION

Variables in a system sometimes follow an exponential relationship as 𝑥 = 𝑒 𝑤 .
If the exponent is a random variable, say 𝑊, 𝑋 = 𝑒 𝑊 is a random variable and the
distribution of 𝑋 is of interest.
An important special case occurs when 𝑊 has a normal distribution. In that case, the
distribution of X is called a lognormal distribution. The name follows from the
transformation 𝑙𝑛𝑋 = 𝑊.
That is, the natural logarithm of 𝑋 is normally distributed.
Probabilities for 𝑋 are obtained from the transformation to 𝑊, but we need to recognize
that the range of 𝑋 is (0, ∞).
Suppose that 𝑊 is normally distributed with mean 𝜃 and variance 𝜔2 ; then the cumulative
distribution function for X is
𝑊
ln𝑥 − 𝜃 ln𝑥 − 𝜃
𝐹 (𝑥 ) = 𝑃(𝑋 ≤ 𝑥 ) = 𝑃(𝑒 ≤ 𝑥 ) = 𝑃(𝑊 ≤ ln𝑥 ) = 𝑃 (𝑍 ≤ ) = Φ( )
𝜔 𝜔
for 𝑥 > 0, where 𝑍 is a standard normal random variable.
Also, 𝐹 (𝑥 ) = 0, for 𝑥 ≤ 0.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 95

The probability density function of X can be obtained from the derivative of F(x).
This derivative is applied to the last term in the expression for F(x), the integral of the
standard normal density function.
Furthermore, from the probability density function, the mean and variance of X can be
derived.
The details are omitted, but a summary of results follows.

Let W have a normal distribution mean 𝜃 and variance 𝜔2

Then 𝑋 = 𝑒 𝑊 is a lognormal random variable with probability density function
1 −(ln𝑥−𝜃 )2
𝑓 (𝑥 ) = ∙ 𝑒 2𝜔2 , 𝑥>0 (4 − 22)
𝑥 ∙ 𝜎 ∙ √2𝜋
The mean and variance of X are
𝜔2
𝜇 = 𝐸 (𝑋) = 𝑒 𝜃+ 2
and
2 2
𝜎 2 = 𝑉(𝑋) = 𝑒 2𝜃+𝜔 (𝑒 𝜔 − 1) (4 − 25)
On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 96
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 97

The lifetime of a product that degrades over time is often modeled by a lognormal random
variable.
For example, this is a common distribution for the lifetime of a semiconductor laser.
A Weibull distribution can also be used in this type of application, and with an appropriate
choice for parameters, it can approximate a selected lognormal distribution.
However, a lognormal distribution is derived from a simple exponential function of a
normal random variable, so it is easy to understand and easy to evaluate probabilities.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 98

EXAMPLE 4-26
The lifetime of a semiconductor laser has a lognormal distribution with 𝜃 = 10 hours
and 𝜔 = 1.5 hours.
What is the probability the lifetime exceeds 10,000 hours?
From the cumulative distribution function for 𝑋
𝑃(𝑋 > 10,000) = 1 − 𝑃(𝑒 𝑊 ≤ 10,000) = 1 − 𝑃(𝑊 ≤ ln 10,000) =
ln 10,000 − 10
= 1 − Φ( ) = 1 − Φ(−0.52) = 1 − 0.3 = 0.7
1.5
What lifetime is exceeded by 99% of lasers?
The question is to determine 𝑥 such that 𝑃(𝑋 > 𝑥 ) = 0.99.
Therefore,
𝑊
ln 𝑥 − 10
0.99 = 𝑃(𝑋 > 𝑥 ) = 𝑃(𝑒 > 𝑥 ) = 𝑃(𝑊 > ln 𝑥 ) = 1 − Φ ( )
1.5
ln 𝑥 − 10 ln 𝑥 − 10
Φ( ) = 0.01 ⟹ = −2.33 ⟹ 𝑥 = 𝑒 6.505 = 668 hours
1.5 1.5
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 99

Determine the mean and standard deviation of lifetime.

Now,
𝜔2
𝐸 (𝑋) = 𝑒 𝜃+ 2 = 𝑒 11.125 = 67,846
2 2
𝑉 (𝑋) = 𝑒 2𝜃+𝜔 (𝑒 𝜔 − 1) = 𝑒 22.25 ∙ (𝑒 2.25 − 1) = 39,070,059,886
so the standard deviation of 𝑋 is 197,661 hours.
Notice that the standard deviation of lifetime is large relative to the mean.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 100

Selected exercises
Exercise 4-6.
The probability density function of the time to failure of an electronic component in a
copier (in hours) is
1 −1
𝑓(𝑥 ) = ∙ 𝑒 1000∙𝑥 , 𝑥 > 0
1000
Determine the probability that
a) A component lasts more than 3000 hours before failure.
b) A component fails in the interval from 1000 to 2000 hours.
c) A component fails before 1000 hours.
d) Determine the number of hours at which 10% of all components have failed.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 101

Exercise 4-13.
Determine the cumulative distribution function for the distribution given with the
probability density function
𝑓(𝑥 ) = 𝑒 −𝑥 , 𝑥>0
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 102

Exercise 4-17.
The probability density function of the length of a hinge for fastening a door is
𝑓 (𝑥 ) = 1.25, 74.6 < 𝑥 < 75.4
millimeters.
a) Determine the cumulative distribution function of the length.
b) Use the cumulative distribution function to determine the probability that a length
exceeds 75 millimeters.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 103

Exercise 4-23.
Suppose that the probability density function of 𝑋 is
𝑓 (𝑥 ) = 0.125 ∙ 𝑥, 0<𝑥<4
Determine the mean and variance of 𝑋.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 104

Exercise 4-49.
The line width for semiconductor manufacturing is assumed to be normally distributed
with a mean of 0.5 micrometer and a standard deviation of 0.05 micrometer.
a) What is the probability that a line width is greater than 0.62 micrometer?
b) What is the probability that a line width is between 0.47 and 0.63 micrometer?
c) The line width of 90% of samples is below what value?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 105

Exercise 4-59.
The diameter of the dot produced by a printer is normally distributed with a mean
diameter of 0.002 inch and a standard deviation of 0.0004 inch.
a) What is the probability that the diameter of a dot exceeds 0.0026 inch?
b) What is the probability that a diameter is between 0.0014 and 0.0026 inch?
c) What standard deviation of diameters is needed so that the probability in part (b) is
0.995?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 106

Exercise 4-61.
Suppose that 𝑋 is a binomial random variable with 𝑛 = 200 and 𝑝 = 0.4
a) Approximate the probability that 𝑋 is less than or equal to 70.
b) Approximate the probability that 𝑋 is greater than 70 and less than 90.
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 107

Exercise 4-80.
The time between the arrival of electronic messages at your computer is exponentially
distributed with a mean of two hours.
a) What is the probability that you do not receive a message during a two-hour period?
b) If you have not had a message in the last four hours, what is the probability that you do
not receive a message in the next two hours? (conditional probability)
c) What is the expected time between your fifth and sixth messages?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 108

Exercise 4-81.
The time between arrivals of taxis at a busy intersection is exponentially distributed with
a mean of 10 minutes.
a) What is the probability that you wait longer than one hour for a taxi?
b) Suppose you have already been waiting for one hour for a taxi, what is the probability
that one arrives within the next 10 minutes? (conditional probability)

1 1
𝜆= =
𝐸(𝑋) 10
𝑥
−
𝐹 (𝑥 ) = 1 − 𝑒 10 , 𝑥>0

60
−
a, 𝑃(𝑋 > 60) = 1 − 𝐹 (60) = 1 − (1 − 𝑒 10 ) = 𝑒 −6 = 0.00248
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 109

A
B
C AB

0 10 60 70
b, Let A: 60 < 𝑋, B: 𝑋 < 70. Then 𝐴 ∩ 𝐵: 60 < 𝑋 < 70
Question is the conditional probability 𝑃(𝐵|𝐴) = 𝑃(𝑋 < 70|60 < 𝑋)
𝑃(𝐴 ∩ 𝐵) 𝑃(60 < 𝑋 < 70 )
𝑃(𝑋 < 70|60 < 𝑋) = 𝑃(𝐵|𝐴) = =
𝑃(𝐴) 𝑃(60 < 𝑋)
𝑃(60 < 𝑋 < 70 ) = 𝐹 (70) − 𝐹 (60) = (1 − 𝑒 − 7 ) − (1 − 𝑒 − 6 ) = 𝑒 −6 − 𝑒 −7 = 0.001567
𝑃(60 < 𝑋 ) = 1 − 𝐹 (60) = 1 − (1 − 𝑒 − 6 ) = 𝑒 −6 = 0.00248
𝑃(60 < 𝑋 < 70 ) 0.001567
𝑃(𝑋 < 70|60 < 𝑋) = = = 0.63211
𝑃(60 < 𝑋) 0.002479

Compare with 𝑃(𝑋 < 10 ) = 𝐹 (10) = 1 − 𝑒 −1 = 0.63211 (lack of memory property)

 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 110

Exercise 4-85.
The lifetime of a mechanical assembly in a vibration test is exponentially distributed with
a mean of 400 hours.
a) What is the probability that an assembly on test fails in less than 100 hours?
b) What is the probability that an assembly operates for more than 500 hours before
failure?
c) If an assembly has been on test for 400 hours without a failure, what is the probability
of a failure in the next 100 hours?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 111

Exercise 4-111.
Assume that the life of a roller bearing follows a Weibull distribution with parameters
𝛽 = 2 and 𝛿 = 10,000 hours.
a) Determine the probability that a bearing lasts at least 8,000 hours.
b) Determine the mean time until failure of a bearing.
c) If 10 bearings are in use and failures occur independently, what is the probability that
all 10 bearings last at least 8,000 hours?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 112

Exercise 4-115.
The life (in hours) of a magnetic resonance imagining machine (MRI) is modeled by a
Weibull distribution with parameters 𝛽 = 2 and 𝛿 = 500 hours.
a) Determine the mean life of the MRI.
b) Determine the variance of the life of the MRI.
c) What is the probability that the MRI fails before 250 hours?
 On the basis of Applied Statistics and Probability for Engineers by D. C. Montgomery - 113

Sample Questions – Chapter 4

Define the probability density function a continuous random variable
Define the cumulative distribution function of a continuous random variable
Give the expected value and the variance of a continuous random variable
Give the probability density function of the continuous uniform random variable
Give the probability density function of a normal random variable
Give the probability density function of an exponential random variable
Give the cumulative distribution function of an exponential random variable