IE 230 Seat # ________ Name ___ < KEY > ___

Please read these directions.

Closed book and notes. 60 minutes.

Covers through the normal distribution, Section 4.6 of Montgomery and Runger, fourth edition.

Cover page and four pages of exam.

Pages 8 and 12 of the Concise Notes.
A normal-distribution cdf table.
No calculator. No need to simplify beyond probability concepts.
For example, unsimplified factorials, integrals, sums, and algebra receive full credit.
Throughout, f denotes probability mass function or probability density function
and F denotes cumulative distribution function.
For one point of credit, write your name neatly on this cover page. Circle your family name.
For one point of credit, write your name on each of pages 1 through 4.

Score ___________________________

Exam #2, October 19, 2010 Schmeiser

IE 230 — Probability & Statistics in Engineering I Name _______________________

Closed book and notes. 60 minutes.

1. Suppose that the random variable X has the discrete uniform distribution over the set
{0, 1,..., 10} and that the random variable Y has the continuous uniform distribution over
the set [0, 10].

(a) (3 pt) T ← F E(X ) = E(Y ).

(b) (3 pts) T F← V(X ) = V(Y ).

(c) (3 pts) T F← FX (5) = FY (5).

(d) (3 pts) T F← f X (5) = f Y (5).

2. (from Montgomery and Runger, 3–113) In 1898 L. J. Bortkiewicz published a book

entitled The Law of Small Numbers. He used data collected over twenty years to show
that the number of soldiers killed by horse kicks each year in each corps in the Prussian
cavalry followed a Poisson distribution with a mean of 0.61.

(a) (7 pts) For a particular corps over one year, determine the value of the probability of
more than one death.
____________________________________________________________
Deaths occur with rate λ = 0.61 deaths per year per corps.
Let N = "number of deaths in one year in a particular corps".
Then N is Poisson with rate µ = λt = (0.61)(1) = 0.61 deaths.
Therefore, P(N > 1) = 1 − [P(N = 0) + P(N = 1)]
−µ 0 −µ 1
R
e µ e µ H
−0.61
=1− J hhhhh + hhhhh J
=1−e [1 + 0.61] ←
Q 0! 1! P

____________________________________________________________

(b) (7 pts) For a particular corps over five years, determine the value of the probability of
no death.
____________________________________________________________
Deaths still occur with rate λ = 0.61 deaths per year per corps.
Now let N denote the number of deaths during five years.
Then N is Poisson with mean µ = λt = (0.61)(5) = 3.05.
−µ 0
e µ −3.05
Therefore, P(N = 0) = hhhhh
=e ←
0!
____________________________________________________________

Exam #2, October 19, 2010 Page 1 of 4 Schmeiser

IE 230 — Probability & Statistics in Engineering I Name _______________________

3. (from Montgomery and Runger, 4–61) The lifetime of a semiconductor laser at a constant
power is normally distributed with a mean of 6000 hours and standard deviation of 500
hours.

(a) (8 pts) Sketch the corresponding normal pdf. Label and scale both axes.
____________________________________________________________
Sketch two axes.
Label the horizontal axis with x and the vertical with f X (x ).
Scale the horizontal axis with at least two numbers.
The easiest is to place the mean 6000 at the center.
Maybe place 5500 and 6500 at the points of inflection.
Maybe place 4500 and 7500 where the bell curve disappears.
Scale the vertical axis.
Placing zero at the bottom of the bell curve is an easy number.
The height at the mode is 1 / (500 √d2π)
dd .

Or sketch a rectangle to determine a height.

____________________________________________________________

(b) (8 pts) Determine the value of the probability that a randomly selected laser fails
between 4000 and 5000 hours. Provide at least three digits of precision.
____________________________________________________________

P(4000 < X < 5000) = P(7000 < X < 8000) = FX (8000) − FX (7000)
= FZ (4)−FZ (2) ≈ 0.999968 − 0.977250 ≈ 0.0227 ←

Notice that including, or excluding, 4000 and 5000 makes no difference.

____________________________________________________________

(c) (8 pts) Let q denote the correct answer to Part (b). If three lasers are used in a
product, and all have independent lifetimes, determine the value of the probability
that all three lasers fail between 4000 and 5000 hours.
____________________________________________________________

Let Li denote that laser i has a lifetime in (4000, 5000).

Then P(Li ) = q ≈ 0.0227.
Then L 1, L 2, and L 3 are Bernoulli trials, with probability of success q .
Let X denote the number of trials that succeed.
So X is binomial with n = 3 and p = q . From Page 8 of the Concise Notes,
I
3M 3 3−3 3
P(N = 3) = f X (3) = J J q (1 − q ) =q ←
L3 O

____________________________________________________________

Exam #2, October 19, 2010 Page 2 of 4 Schmeiser

IE 230 — Probability & Statistics in Engineering I Name _______________________

4. Result. For c = 1, 2,..., the geometric cdf is FX (c ) = 1 − (1 − p )c , where p denotes the

probability of success on each Bernoulli trial.
(3 pts each) For each of the six lines (a–f), state why the corresponding equality is true.

Each blank requires one reason; reasons may be reused. Choose the reasons from this
list:
(i) Events partition the sample space.
(ii) Events are complementary.
(iii) Events are mutually exclusive.
(iv) Events are independent.
(v) Events are the same.
(vi) Definition of conditional probability.
(vii) Definition of cumulative distribution function.
(viii) Definition of probability mass function.
(ix) Multiplication Rule.
(x) Total Probability.
(xi) Substitute known values.

Proof. Let c be a positive integer. Let Ai denote success on the i th trial.

FX (c ) = P(X ≤ c ) (a) ___ < vii > ___

= 1 − P(X > c ) (b) ___ < ii > ___

= 1 − P(A ′1 ∩ A ′2 ∩ . . . ∩ A ′c ) (c) ___ < v > ___

= 1 − P(A ′1) P(A ′2) . . . P(A ′c ) (d) ___ < iv > ___

= 1 − [1 − P(A 1)] [1 − P(A 2)] . . . [1 − P(Ac )] (e) ___ < ii > ___

c
= 1 − (1 − p ) (f) ___ < xi > ___

Exam #2, October 19, 2010 Page 3 of 4 Schmeiser

IE 230 — Probability & Statistics in Engineering I Name _______________________

5. For (a–e), provide the name of the corresponding family of distributions.

(a) (3 pts) The number of coin flips until the fourth "head".
____________________________________________________________
negative binomial
____________________________________________________________
(b) (3 pts) When drawing, without replacement, five cards from a standard card deck,
the number of aces drawn.
____________________________________________________________
hypergeometric ←
____________________________________________________________
(c) (3 pts) When drawing, with replacement, five cards from a standard card deck, the
number of aces drawn.
____________________________________________________________
binomial ←
____________________________________________________________
(d) (3 pts) The number of insect parts in a candy bar.
____________________________________________________________
Poisson ←
____________________________________________________________
(e) (3 pts) When playing Keno, the number of matched numbers.
____________________________________________________________
hypergeometric ←
____________________________________________________________
6. Short answer.
I
6M
(a) (3 pts) Evaluate J J, the number of ways to choose two items from six items.
L2 O
____________________________________________________________
I
6 M hhhhh
6!
J J = = 15 ←
L2 O 2! 4!
____________________________________________________________
(b) (3 pts) What is the numerical value of 0! ?
____________________________________________________________
1! = 1 ←
____________________________________________________________
(c) (3 pts) T ← F P(π < X ≤ 2π) = FX (2π) − FX (π).

(d) (3 pts) The mean of the binomial distribution is n p , the number of trials multiplied
by the probability of success. The mean is in units of "trials".
What are the units of the variance?
____________________________________________________________
"trials" squared ←
____________________________________________________________
(e) (3 pts) The mean of a distribution corresponds to the center of gravity.
The variance of a distribution corresponds to...
____________________________________________________________
the moment of inertia. ←
____________________________________________________________

Exam #2, October 19, 2010 Page 4 of 4 Schmeiser

IE 230 — Probability & Statistics in Engineering I Name _______________________

Discrete Distributions: Summary Table

i
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
i
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

random distribution range probability expected variance

variable
i
name mass function value
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
i
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

n n

Σ xi f (xi ) Σ (xi − µ)
2
X general x 1, x 2, . . . , xn P(X = x ) f (xi )
i =1 i =1
2 2
= f (x ) = µ = µX = σ = σX
= f X (x ) = E(X ) = V(X )
2 2
i
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
= E(X ) − µ
n n

Σ xi / n [ Σ xi / n ] − µ
2 2
X discrete x 1, x 2, . . . , xn 1/n
i =1 i =1
i uniform
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
2 2
a +b c (n −1)
X equal-space x = a ,a +c ,...,b 1/n hhhh h
hhhhhhhh

2 12
i uniform where n = (b −a +c ) / c
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
x 1−x
"# successes in indicator x = 0, 1 p (1−p ) p p (1−p )
1 Bernoulli variable
i trial" where p = P("success")
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
n x n −x
"# successes in binomial x = 0, 1,..., n Cx p (1−p ) np np (1−p )
n Bernoulli
i trials" where p = P("success")
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

K N −K N (N −n )
"# successes in hyper- x = Cx Cn −x / Cn np np (1−p ) hhhhhh
+ (N −1)
a sample of geometric (n −(N −K )) ,
size n from ..., min{K , n } where p =K /N
a population (sampling and
of size N without integer
containing replacement)
i K successes"
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
x −1 2
"# Bernoulli geometric x = 1, 2,... p (1−p ) 1/p (1−p ) / p
trials until
i 1st success" where p = P("success")
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
x −1 r x −r 2
"# Bernoulli negative x = r , r +1,... Cr −1 p (1−p ) r /p r (1−p ) / p
trials until binomial
i r th success" where p = P("success")
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
−µ x
"# of counts in Poisson x = 0, 1,... e µ / x! µ µ
time t from a
Poisson process where µ = λt
i
i
with rate λ"
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

x
Result. For x = 1, 2,..., the geometric cdf is FX (x ) = 1 − (1 − p ) .
Result. The geometric distribution is the only discrete memoryless distribution.
That is, P(X > x + c | X > x ) = P(X > c ).
Result. The binomial distribution with p = K / N is a good approximation to the hypergeometric
distribution when n is small compared to N .

Purdue University – 5 of 22 – B.W. Schmeiser

IE230 CONCISE NOTES Revised August 25, 2008

Continuous Distributions: Summary Table

iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

random distribution range cumulative probability expected variance

variable
i
name distrib. func. density func. value
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

∞ ∞
J
dF (y ) 2
X general (−∞, ∞) P(X ≤ x ) hhhhhh

dy
J
J
y =x ∫ xf (x )dx ∫ (x −µ) f (x )dx
−∞ −∞
2 2
= F (x ) = f (x )
= σ = σX = µ = µX
= FX (x ) = f X (x )
= V(X ) = E(X )
2 2
= E(X ) − µ
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

2
x −a
hhhhh
1
hhhhh
a +b
hhhhh
(b − a )
h
hhhhhhh
X continuous [a , b ]
b −a b −a 2 12
uniform
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

2
2(x − d )
(x −a )f (x ) / 2 hhhhhhhhhhhh a +m +b
hhhhhhh
(b −a ) − (m −a )(b −m )
h hhhhhhhhhhhhhhhhhh
X triangular [a , b ]
if x ≤ m , else (b −a )(m −d ) 3 18
1−(b −x )f (x )/2 (d = a if x ≤ m , else d = b )
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
2
−1 I hxhhh
hhh
−µ M
J J
2 L σ O

e 2
sum of normal (−∞, ∞) Table III hhhhhhhhh
µ σ
dd σ
√d2π
random (or
variables Gaussian)
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

−λx −λx 2
time to exponential [0, ∞) 1 − e λe 1/λ 1/λ
Poisson
count 1
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

∞ −λx k r r −1 −λx
e (λx ) λ x e
Σ
2
time to Erlang [0, ∞) hhhhhhhhh h
hhhhhhhhh
r /λ r /λ
k =r
k! (r − 1)!
Poisson
count r
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

r r −1 −λx
λ x e 2
lifetime gamma [0, ∞) numerical h
hhhhhhhhh
r /λ r /λ
Γ(r )
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

β
β−1 −(x /δ)
−(x /δ)
β 1 βx e
2 2 2
lifetime Weibull [0, ∞) 1−e δΓ(1+ hh ) δ Γ(1+ hh )−µ
h
hhhhhhhhhh
β β β δ
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii
iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii

∞
r −1 −x
Definition. For any r > 0, the gamma function is Γ(r ) = ∫ x e dx .
0
Result. Γ(r ) = (r − 1)Γ(r − 1). In particular, if r is a positive integer, then Γ(r ) = (r −1)!.
Result. The exponential distribution is the only continuous memoryless distribution.
That is, P(X > x + c | X > x ) = P(X > c ).
Definition. A lifetime distribution is continuous with range [0, ∞).
Modeling lifetimes. Some useful lifetime distributions are the exponential, Erlang, gamma, and
Weibull.

Purdue University – 12 of 22 – B.W. Schmeiser