You are on page 1of 11

Goble05.

book Page 245 Thursday, March 31, 2005 10:39 PM

Appendix B
Probability

Probability Assignment
Probability is a quantitative method of expressing chances. A probability
is assigned a number between zero and one, inclusive. A probability
assignment of zero means that the event is never expected. A probability
assignment of one means that the event is always expected.
Probabilities are often assigned based on historical “frequency of
occurrence.” An experiment is repeated many times, say N. A quantity is
tabulated for each possible outcome of the experiment. For any particular
outcome, the probability is determined by dividing the number of
occurrences, n, by the number of trials.

n
P(E) = (B-1)
N
The values become more certain as the number of trials is increased. A
definition of probability based on this concept is stated in Equation B-2:

n
P( E ) =
lim N (B-2)
N →∞

Venn Diagrams
A convenient way to depict the outcomes of an experiment is through the
use of the Venn diagram. These diagrams were created by John Venn
(1834-1923), an English mathematician and cleric. They provide visual
representation of data sets, including experimental outcomes. The

245
Goble05.book Page 246 Thursday, March 31, 2005 10:39 PM

246 Appendix B: Probability

diagrams are drawn by using the area of a rectangle to represent all


possible outcomes; this area is known as the “sample space.” Any
particular outcome is shown by using a portion of the area within the
rectangle.

For the toss of a fair pair of dice, the possible outcomes are shown in the
Venn diagram of Figure B-1. The outcomes do not occupy the same area on
the diagram. The probabilities of some outcomes are more likely than
others; these occupy more area. For example, the area occupied by an
outcome of “2” is 1/36 of the total. The area occupied by the outcome “7”
is 6/36 of the total. Again, the area occupied by each outcome is
proportional to its probability.

2 3 4
5
6
7
8
9
10 11 12
Figure B-1. Venn Diagram - Dice Toss

A Venn diagram is often used to identify the attributes of possible


outcomes. Outcomes are grouped into sets that are based on some
characteristic or combination of characteristics. The graphical nature of the
Venn diagram is especially useful in showing these combinations of sets:
unions, intersections, and complementary events.

A Union of some number of sets (A, B, C) is defined as any event in either


set A or set B or set C. This is represented in a Venn diagram as shown in
Figure B-2. Vertical lines extend through three circles, the A circle, the B
circle, and the C circle.

An Intersection of sets (A, B, C) is defined as any event in sets A and B and


C. This is represented in a Venn diagram as shown in Figure B-3. Only one
small area is in all three circles -- that area is marked with lines.

Complementary sets are easily shown on Venn diagrams. Since the diagram
represents the entire sample space, all area not enclosed within an event is
Goble05.book Page 247 Thursday, March 31, 2005 10:39 PM

Appendix B: Probability 247

A B
C
C
Figure B-2. Union of Sets A, B, and C

A B

C
Figure B-3. Intersection of Sets A, B, and C

the complement of the event. In Figure B-4, a circle represents the set A. Its
complement is set B, represented by the remainder of the diagram.

Mutually exclusive sets are defined as sets that cannot happen at the same
time. Mutually exclusive event sets are easily recognized on a Venn
diagram. In Figure B-5, the event sets A and B are shown. There is no
common space within the A circle and the B circle. There is no intersection
between A and B. They cannot happen at the same time and are, therefore,
mutually exclusive.
Goble05.book Page 248 Thursday, March 31, 2005 10:39 PM

248 Appendix B: Probability

Figure B-4. Complementary Events

Figure B-5. Mutually Exclusive Events

Combining Probabilities
Certain rules help to calculate the probability of combinations of events.
Combinations of events are common in the field of reliability evaluation.
Often system failures occur only when certain combinations of events
happen during certain times.

Independent Events

If the occurrence of an event from set A does not affect the probability of
events from set B, then sets A and B are defined to be “independent”; for
example, the outcome of one coin toss does not affect the next toss. The
Goble05.book Page 249 Thursday, March 31, 2005 10:39 PM

Appendix B: Probability 249

outcome of one coin toss does not give us any information about the
outcome of a subsequent independent coin toss. When two event sets are
independent, the probability of getting an event from set A and set B (the
intersection) is given by the formula:

P(A ∩ B) = P(A) x P(B) (B-3)

Independent events are different than mutually exclusive events. Consider


two events, A and B that are mutually exclusive. Knowing that A has
occurred tells us that B cannot occur. If events A and B are independent,
knowing that A occurred tells us nothing about B. Two events, A and B,
cannot be both mutually exclusive and independent.

EXAMPLE B-1

Problem: Two fair coins are flipped into the air. What is the probability that
both coins will land with heads showing?

Solution: Each coin toss has only two possible outcomes: heads or tails.
Each outcome has a probability of one half. The coin tosses are independent.
Therefore,

P(two heads) = P (head 1) × P (head 2) = 1/2 × 1/2 = 1/4

EXAMPLE B-2

Problem: A pair of fair (well balanced) dice is rolled. What is the probability of
getting “snake eyes” -- one dot on each die?

Solution: The outcome of one die does not affect the outcome of the other
die. Therefore, the events are independent. The probability of getting one dot
can be obtained by noting that there are six sides on the die and that each
side is equally likely. The probability of getting one dot is one sixth (1/6). The
probability of getting “snake eyes” is represented as:

P(1,1) = 1/6 × 1/6 = 1/36

Check the area occupied by the “2” result on Figure B-1. Is that area equal to
one thirty-sixth?

Probability Summation

If the probability of getting a result from set A equals 0.2 and the
probability of getting a result from set B equals 0.3, what is the probability
of getting a result from either set A or set B?
Goble05.book Page 250 Thursday, March 31, 2005 10:39 PM

250 Appendix B: Probability

EXAMPLE B-3

Problem: A controller fails only if the input power fails and the controller
battery fails. Assume that these factors are independent. For a time interval of
five years, the probability of input power failure is 0.0001 and the probability of
battery failure is 0.01. What is the probability of controller failure during the
time interval of five years?

Solution: Since input power and battery failure are independent, Equation
B-3 gives the probability of both events:

P(Controller Failure) = 0.0001 × 0.01 = 0.000001

It would be natural to assume that the answer is 0.5, the sum of the above
probabilities, but that answer is not always correct. Look at the Venn
diagram in Figure B-6. If the area of set A (6/36) is added to the area of set
B (6/36), the answer (12/36) is too large. (The answer should be 11/36.)
Since there is an intersection between sets A and B, the area of the
intersection has been counted twice. When summing probabilities, the
intersections must be subtracted. Thus, the probability of the union of
event sets A and B is given by:

P( A ∪ B ) = P( A ) + P( B ) - P( A ∩ B ) (B-4)

If set A and set B are mutually exclusive so there is no intersection, then


the following can be stated.

P( A ∪ B ) = P( A ) + P( B ) (B-5)

EXAMPLE B-4

Problem: A pair of fair dice is rolled. What is the probability of getting a sum
of seven?

Solution: A sum of seven dots on the dice can be obtained in a number of


different ways; these are described by the sets {1,6}, {2,5}, {3,4}, {4,3}, {5,2},
and {6,1}. Each specific combination has a probability of 1/36. The
combinations are mutually exclusive; therefore, Equation B-5 can be used.

P(seven dots) = 1/36 + 1/36 + 1/36 + 1/36 + 1/36 + 1/36 = 1/6

Conditional Probability
Often it is required to calculate the probability of some event under
specific circumstances. The probability of event A, given that event B has
occurred, may need to be calculated. Such a probability is called a
Goble05.book Page 251 Thursday, March 31, 2005 10:39 PM

Appendix B: Probability 251

EXAMPLE B-5

Problem: A pair of fair dice is rolled. What is the probability of getting an even
number on both dice?

Solution: On each die there are six numbers. Three of the numbers are odd
(1, 3, 5) and three of the numbers are even (2, 4, 6). All numbers are mutually
exclusive. Equation B-5 gives the probability of getting an even number on
one die.

P(even) = P(2,4,6) = P(2) + P(4) + P(6) = 1/6 + 1/6 + 1/6 = 1/2

The outcome of one die is independent of the other die. Therefore,

P(even, even) = P(Set A even) × P(Set B even) = 1/2 × 1/2 = 1/4

EXAMPLE B-6

Problem: A pair of fair dice is rolled. What is the probability of getting two dots
on either or both dice?

Solution: The probability of getting two dots on die A or B equals 1/6. The
probability of getting two dots on both dice though is 1/36. Because these
events of independent, they are not mutually exclusive. Therefore, we can use
Equation B-4.

P(A OR B) = P(A) + P(B) – P(A × B) = 1/6 + 1/6 – 1/36 = 11/36

This is evident in Figure B-6, a Venn diagram of the problem.

“conditional probability.” The situation can be envisioned by examining


the Venn diagram in Figure B-7.

Normally, the probability of event A would be given by the area of circle A


divided by the total area. Conditional probability is different. For example,
consider the situation when Event B has occurred. This means that only
the state space within the area of circle B needs to be examined. This is a
substantially reduced area! The desired probability is the area of circle A
within circle B, divided by the area of circle B, expressed by:

P( A ∩ B )
P( A|B ) = (B-6)
P( B )
This reads: the probability of A, given B, is equal to the probability of the
intersection of A and B divided by the probability of B. The area of circle A
within circle B represents the probability of the intersection of A and B.
The area of circle B equals the probability of B.
Goble05.book Page 252 Thursday, March 31, 2005 10:39 PM

252 Appendix B: Probability

Event B Area = 1/6

Event A
Area = 1/6

Figure B-6. Venn Diagram of Example B-6

A B

Figure B-7. Venn Diagram for Conditional Probability

The formula for conditional probability can be rearranged as:

P(A ∩ B) = P(A | B) x P(B) (B-7)

This states that the intersection of events A and B can be obtained by


multiplying the probability of A, given B, times the probability of B. When
the statistics are kept in a conditional format, this equation can be useful.
Goble05.book Page 253 Thursday, March 31, 2005 10:39 PM

Appendix B: Probability 253

EXAMPLE B-7

Problem: A pair of fair dice is rolled. What is the probability of getting a two on
both dice given that one die has a two?

Solution: The probability of {2,2}, given that one die has a two, is given by
Equation B-6:

P(2,2) = 1/36 / 1/6 = 1/6

In this case, the answer is intuitive since the outcome of each die is
independent.

EXAMPLE B-8

Problem: A pair of fair dice is rolled. What is the probability of getting a sum
of seven, given that exactly one die shows a two?

Solution: There are only two ways to get a sum of seven, given that one die
has a two. Those two combinations are {2,5} and {5,2}. There are 10
combinations that show a two on exactly one die. These sets are {2,1}, {2,3},
{2,4}, {2,5}, {2,6}, {1,2}. {3,2}, {4,2}, {5,2}, and {6,2). Using Equation B-6:

P( A ∩ B )
P( A | B ) =
P( B )
2
= 36
10
36
2
=
10
This is shown graphically in Figure B-8.

2,1 2,3 2,4 2,5 2,6

1,2 3,2 4,2 5,2 6,2

Figure B-8. Probability Diagram of Example B-8


Goble05.book Page 254 Thursday, March 31, 2005 10:39 PM

254 Appendix B: Probability

Bayes' Rule
Consider an event A. The state space in which it exists is divided into two
mutually exclusive sections, B and B' (Figure B-9). Event A can be written
as:

A = ( A ∩ B ) ∪ ( A ∩ B′ ) (B-8)

Since AB and AB' are mutually exclusive,

P( A ) = P( A ∩ B ) + P( A ∩ B ′ ) (B-9)

Substituting Equation B-7 into B-9,

P(A) = P(A | B) x P(B) + P(A | B′) x P(B′) (B-10)

This states that the probability of event A equals the conditional


probability of A, given that B has occurred, plus the conditional
probability of A, given that B has not occurred. This is known as Bayes'
rule. It is used in many aspects of reliability engineering.

B B’
Figure B-9. Event A Partitioned
Goble05.book Page 255 Thursday, March 31, 2005 10:39 PM

Appendix B: Probability 255

EXAMPLE B-9

Problem: The workday is divided into three mutually exclusive time periods:
day shift, evening shift, and night shift. Day shift lasts ten hours. Evening shift
is eight hours. Night shift is six hours. Logs show that in the last year (8760
hours) one failure occurred during the day shift (one failure in 3650 hours),
two failures occurred during the evening shift (two failures in 2920 hours), and
seven failures occurred during the night shift (seven failures in 2190 hours).
What is the overall probability of failure?

Solution: Define event A as failure. Define event B1 as the day shift, B2 as


the evening shift, and B3 as the night shift. The probability of failure given,
event B1 (day shift) is calculated knowing that one failure occurred in 3650
hours (one third of the hours in one year). A variation of Equation B-10 can be
used where P(B1) is day shift probability, P(B2) is the evening shift probability,
and P(B3) is the night shift probability.

P(fail) = P(fail | B1) × P(B1) + P(fail | B2) × P(B2) + P(fail | B3) × P(B3)

The probabilities of failure for each shift are calculated by dividing the number
of failures during each shift by the numbers of hours in each shift. Substituting
the numbers:

P(fail) = (1/3650 × 10/24) + (2/2920 × 8/24) + (7/2190 × 6/24)

= 0.000114 + 0.000226 + 0.000799 = 0.001139

EXAMPLE B-10

Problem: A company manufactures controllers at two locations. Sixty percent


are manufactured in plant X. Forty percent are manufactured in plant Y.
Controllers manufactured in plant X have a 0.00016 probability of failure in a
one year period. Controllers manufactured in plant Y have a 0.00022
probability of failure in one year. A purchased controller can come randomly
from either source. What is the probability of a controller failure?

Solution: Define controller failure as event A. Define event B1 as plant X


manufacture. Define event B as plant Y manufacture. Using Equation B-10,
substitute the values to obtain:

P(fail) = (0.00016 × 0.6) + (0.00022 × 0.4) = 0.000096 + 0.000088 = 0.000184

You might also like