7/30/2019 nspch6
1/48
Statistical methods
1. Pareto diagrams and dot diagrams:
It is essential to collect data to provide the vital information
necessary to solve engineering problems. Graphicalillustration is an effective tool to describe the information.Pareto diagrams and dot diagrams fall into this category.
Example 1:
A computer-controlled lathe whose performance wasbelow par, workers recorded the following causes andtheir frequencies.
Power fluctuations 6
Controller not stable 22Operator error 13
Worn tool not replaced 2
Other 5
7/30/2019 nspch6
2/48
The data is presented asa special case of barchart called a Paretodiagram shown in Fig.1.This diagram showsParetos empirical lawthat any assortment of
events consists of a fewmajor and many minorelements. Typically, twoor three elements will
account for more thanhalf of the totalfrequency.
7/30/2019 nspch6
3/48
The figure visually emphasizes the importance of reducing
the frequency of controller misbehavior.
As a second step toward improvement of the process, data
were collected on the deviations of cutting speed from the
target value set by the controller. The seven observed
values of (cutting speed)-(target)3 6 -2 4 7 4 3
are plotted as a dot diagram shown in Fig.2. The dot diagram
visually summarizes the information that the lathe is,
generally, running fast.
7/30/2019 nspch6
4/48
It helps to develop efficient experimental designs and
methods for identifying primary causal factors that
contribute to the variability in a response such as cuttingspeed.
7/30/2019 nspch6
5/48
2. Frequency distributions:
Properties of frequency distributions relating to their steps are
best exhibited by means of graphs. Most common form is
the histogram. The histogram of a frequency distribution isconstructed of adjacent rectangles, the height of rectangles
represent external between successive boundaries.
Example 2: A histogram reveals the solution to a grinding
operation problem.
A metallurgical engineer was experiencing trouble with a
grinding operation. The grinding action was produced by
pellets. After some thought he collected the sample of
pellets used for grinding, took them home, spread them outof his kitchen table, and measured their diameters. His
histogram displayed in Fig.3. What does the histogram
reveal?
7/30/2019 nspch6
6/48
Solution:
The histogram exhibitstwo distinct peaks, onefor a group of pelletswhose diameters arecentered near 25 andthe other concerned
near 40.By getting his supplier todo a better sort, so allthe pellets would be
essentially from the firstgroup, the engineercompletely solved hisproblem.
7/30/2019 nspch6
7/48
3. Descriptive Measures:
(a) Mean:
Mean is the average defined by
(1)
To emphasize that it is based on a set of observation, it is
often referred as Sample mean , .
(b) Median:
It is a descriptive measure of the centre, or location of aset of data. It is used to eliminate the effect of extreme
(very large or very small) values. The median of n
observations can be defined loosely as the middle most
value when the data is arranged to size. (i.e.)
1
n
i
i
x
xn
==
x
7/30/2019 nspch6
8/48
if nis even
average of observation if nis odd.
Example 3: Calculation of the sample median with even
sample size
An engineering group receives email requests for
technical information from sales and service persons. The
daily numbers for six days were 11,9,17,19,4, and 15. Find
the mean and the median.
Solution: The mean is
requests
1
2
n +
12 2n n ++
11 9 17 19 4 1512.5
6x
+ + + + += =
7/30/2019 nspch6
9/48
and, ordering the data from the smallest to largest
the median, the mean of the third and fourth largest values,is 13 requests.
(c)Sample Variance ( ): It is the average of the squared
deviations from the mean .
(2)
There are (n-1) independent deviations .
4 9 11 15 17 19
2s
x
( )2
2 1
1
n
i
i
x x
sn
=
=
( )ix x
7/30/2019 nspch6
10/48
Example 4: Calculation of sample variance
The delay times (handling, setting, and positioning the
tools) for cutting 6 parts on an engine lathe are 0.6, 1.2,0.9, 1.0, 0.6, and 0.8 minutes. Calculate .
Solution: First we calculate the mean:
Then we set up the work required to find in
the following table:
2s
0.6 1.2 0.9 1.0 0.6 0.80.85
6x
+ + + + += =
( )2
ix x
7/30/2019 nspch6
11/48
We divide 0.2750 by (6 1) = 5 to obtain
2 20.27500.055 (min )
5
s ute= =
7/30/2019 nspch6
12/48
(d) Standard deviation: In the above example, is inwrong limits (minutes)2. Hence, standard deviation of nobservations is defined as the square root
of their variance.
(3)
Example 5: Calculation of sample standard deviation
With reference to the previous example, calculate s.
Solution: From the previous example, . Take thesquare root and get
minutes
2s
1 2
, ,....,n
x x x
( )2
1
1
n
i
i
x x
sn
=
=
2
0.055s =
0.055 0.23s = =
7/30/2019 nspch6
13/48
The standard deviation and the variance are measures of
absolute variation, that is, they measure the actual
amount of variation in a set of data, and they depend on thescale of measurement. To compare the variation in several
sets of data, it is generally desirable to use a measure of
relative variation, for instance, the coefficient of
variation, which gives the standard deviation as apercentage of the mean.
(e)Coefficient of variation:
(4).100s
Vx
=
7/30/2019 nspch6
14/48
Example 6: The coefficient of variation for comparing
relative preciseness
Measurements made with one micrometer of the diameter ofa ball bearing have a mean of 3.92 mm and a standard
deviation of 0.015 mm, whereas measurements made with
another micrometer of the unstretched length of a spring
have a mean of 1.54 inches and a standard deviation of0.008 inch. Which of these two measuring instruments is
relatively more precise?
Solution: For the first micrometer the coefficient of variationis 0.015
.100 0.38%3.92
V = =
7/30/2019 nspch6
15/48
And for the second micrometer the coefficient of variation is
Thus, the measurements made with the first micrometer arerelatively more precise.
4. Permutations and Combinations:
(a) Permutations: The number permutations of r objectsselected from a set of n distinct objects is
(5)
0.008
.100 0.52%1.54V = =
( 1)( 2).......( 1)( )! !
( )! ( )!n r
n n n n r n r nP
n r n r
+ = =
7/30/2019 nspch6
16/48
Example 7: Evaluating a permutation
In how many different ways can one make a first, second,third, and fourth choice among 12 firms leasingconstruction equipment?
Solution: For n= 12 and r= 4, the first formula yields
Example 8: The number of ways to assemble chips in acontroller
An electronic controlling mechanism requires 5 identicalmemory chips. In how many ways can this mechanism beassembled by placing the 5 chips in the 5 positions withinthe controller?
12 4
12! 12! 12.11.10.9.8!11,880
(12 4)! 8! 8!
P = = = =
7/30/2019 nspch6
17/48
Solution: For n= 5 and r= 5, the first formula yields
And the second formula yields
5 5
5.4.3.2.1 120P = =
( )5 55! 5! 5! 120
5 5 ! 0!P = = = =
7/30/2019 nspch6
18/48
(b)Combinations:
Combination of n objects taken r at a time and denoted by
(6a)
(6b)
n r
nC or
r
( 1)( 2).....( 1)
!
n n n n n r
r r
+=
!
!( )!
n nor
r r n r
=
7/30/2019 nspch6
19/48
Example 9: Evaluating a combination
In how many ways different ways can 3 of 20 laboratory
assistants be chosen to assist with an experiment?
Solution: For n= 20 and r = 3, the first formula for
yields
Example 10: Selection of machines for an experiment
A calibration study needs to be conducted to see if the
readings on 15 test machines are giving similar results. In
how many ways can 3 of the 15 be selected for the initialinvestigation?
Solution:
n
r
20 20.19.181,140
3 3!
= =
15 15.14.13455
3 3.2.1
ways
= =
7/30/2019 nspch6
20/48
Note that selecting which 3 machines to use is the same as
selecting which 12 not to include. That is, according to the
second formula
Example 11: The number of choices of new researchers
In how many different ways can the director of a research
laboratory choose 2 chemists from among 7 applicants and
3 physicists from among 9 applicants?
Solution: The 2 chemists can be chosen in ways
and the 3 physicists can be chosen in ways. Bythe multiplication rule, the whole selection can be made in
21 * 84 = 1,764 ways
15 1515! 15!
12 312!3! 3!12!
= = =
721
2
=
9
843
=
7/30/2019 nspch6
21/48
5. Probability:
If there are nequally likely possibilities, of which one must
occur and sare regarded as favorable, or as a success,then the probability of a success is given by .
Example 12: Well-shuffled cards are equally likely to be
selected.What is the probability of drawing an ace from a well-
shuffled deck of 52 playing cards?
Solution: There are s= 4 aces among the n= 52 cards, so
we get
s
n
4 1
52 13
s
n= =
7/30/2019 nspch6
22/48
Example 13: Random selection results in the equallylikely case
If 3 of 20 tires in storage are defective and 4 of them are
randomly chosen for inspection (that is, each tire has thesame chance of being selected), what is the probability thatonly one of the defective tires will be included?
Solution: There are equally likely ways of choosing4 of 20 tires, so n = 4,845. The number of favorableoutcomes is the number of ways in which one of thedefective tires and three of the nondefective tires can be
selected, or
It follows that the probability is
or approximately 0.42
20
4,8454
=
3 173* 680 2,040.
1 3s
= = =
2, 040 8
4,845 19
s
n= =
7/30/2019 nspch6
23/48
(a)The frequency interpretation of probability
The probability of an event (or outcome) is the proportion
of times the event would occur in a long run of repeatedexperiments.
Example 14: Long-run relative frequency approximation
of probability
If records show that 294 of 300 ceramic insulators testedwere able to withstand a certain thermal shock, what is the
probability that any one untested insulator will be able to
withstand the thermal shock?
Solution: Among the insulators tested, were able to
withstand the thermal shock, and we use this figure as an
estimate of the probability.
294 0.98300
=
7/30/2019 nspch6
24/48
(b) Axioms of Probability:
Axiom 1 for each event A in S. (7a)
Axiom 2 (7b)Axiom 3 If A and Bare mutually exclusive events in S, then
(7c)
The first axioms states that probabilities are real numbers
on the interval from 0 to 1. The second axiom states that
the sample space as a whole is assigned a probability of 1
and this expresses the idea that the probability of a certainevent, an event which must happen, is equal to 1. The
third axiom states that probability functions must be
additive.
0 ( ) 1P A
( ) ( ) ( ).P A B P A P B = +
( ) 1P S =
7/30/2019 nspch6
25/48
7/30/2019 nspch6
26/48
(c) General addition rule of probability
Theorem: If A and Bare any events in S, then
(8)
Example 15: Using the general addition rule for
probability
With reference to the lawn-mower-rating example, find theprobability that a lawn mower will be rated easy to operate
or having high average cost of repairs, namely
Solution: Given,
we substitute into the formula of above theorem and get
( ) ( )( ) ( )P A B P A P B P A B = +
( )1 1 .P E C
( ) ( ) ( )1 1 1 10.40, 0.30 0.12P E P C and P E C = = =
( )1 1 0.40 0.30 0.12 0.58P E C = + =
7/30/2019 nspch6
27/48
Example 16: The probability of requiring repair underwarranty.
If the probabilities are 0.87, 0.36, and 0.29 that, while
under warranty, a new car will require repairs on theengine, drive train, or both, what is the probability that acar will require one or the other or both kinds of repairsunder the warranty?
Solution: Substituting these given values into the formula ofabove theorem, we get
(d) Conditional Probability:
If A and B are any events in S and , the
conditional probability of A given Bis
(9)
0.87 0.36 0.29 0.94+ =
( ) 0P B
( )( )
( )
P A BP A B
P B
=
7/30/2019 nspch6
28/48
Example 17: Calculating a conditional probability
If the probability that a communication system will have highfidelity is 0.81 and the probability that it will have highfidelity and high selectivity is 0.18, what is the probabilitythat a system with high fidelity will also have highselectivity?
Solution: If A is the event that a communication system has
high selectivity and B is the event that it has high fidelity,we have
and substitution into the formula yields
( ) ( )0.81 0.18,P B and P A B= =
( )0.18 2
0.81 9P A B = =
7/30/2019 nspch6
29/48
Example 18: The conditional probability that a lawn
mover is easy to operate given that repairs are costly
Referring to the lawn-mower-rating example, for which theprobabilities of the individual outcomes are given in fig 1,
use the results obtained to find
Solution: Since we had
substitution into the formula for a conditional probability
yields
1 1( )P E C
( ) ( )1 1 10.12 0.30,P E C and P C = =
( )1 10.12
0.400.30
P E C = =
7/30/2019 nspch6
30/48
(e) General multiplication rule of probability
Theorem: If A and Bare any events in S, then
(10)
Example 19: Using the general multiplication rule of
probability
The supervisor of a group of 20 construction workers wants
to get the opinion of 2 of them ( to be selected at random)
about certain new safety regulations. If 12 workers favor
the new regulations and the other 8 are against them,
what is the probability that both of the workers chosen by
the supervisor will be against the new safety regulations?
( ) ( ). ( ) ( ) 0( ). ( ) ( ) 0
P A B P A P B A if P A
P B P A B if P B =
=
7/30/2019 nspch6
31/48
Solution: Assuming equal probabilities for each selection
(which is what we mean by the selections being random),
the probability that the first worker selected will be againstthe new safety regulations is , and the probability that
the second worker selected will be against the new safety
regulations given that the first one is against them is
Thus, the desired probability is
(f) Special product rule of probability
Two events A and Bare independent events if and only if
(11)
8
20
719
8 7 14. .20 19 95
=
( ) ( ). ( )P A B P A P B =
7/30/2019 nspch6
32/48
Example 20: The outcome to unrelated parts of an
experiment can be treated as independent
What is the probability of getting two heads in two flips ofa balanced coin?
Solution: Since the probability of heads is for each flip
and the two flips are independent the probability is
Example 21: Independence and selection with and
without replacement
Two cards are drawn at random from an ordinary deck
of 52 playing cards. What is the probability of getting two
aces if
1
2
1 1 1. .2 2 4
=
7/30/2019 nspch6
33/48
(a) the first card is replaced before the second card isdrawn;
(b) The first card is not replaced before the second card is
drawn?Solution:
(a) Since there are four aces among the 52 cards, we get
(b) Since there are only three aces among the 51 cardsthat remain after one ace has been removed from thedeck, we get
Note that so independence is violated when the
sampling is without replacement.
4 4 1. .
52 52 169
=
4 3 1. .
52 51 221=
1 4 4. ,
221 52 52
7/30/2019 nspch6
34/48
Example 22: Checking if two events are independent
under an assigned probability
If are the eventsC and D independent?
Solution: Since and not 0.24,
the two events are not independent.
Example 23: Assigning probability by the special
product rule
Let A be the event that raw material is available when
needed and B be the event that the machining time is less
than 1 hour. If assign
probability to the event
( ) ( ) ( )0.65, 0.40P C P D and P C D= =
( ) ( ) ( )( ). 0.65 0.40 0.26P C P D = =
( ) ( )0.8 0.7,P A and P B= =A B
7/30/2019 nspch6
35/48
Solution: Since the events A and B concern unrelated stepsin the manufacturing process, we invoke independenceand make the assignment
Example 24: The extended special product rule ofprobability
What is the probability of not rolling any 6s in four rolls ofa balanced die?
Solution: The probability is
( ) ( ) ( ) ( ) ( ). 0.8 . 0.7 0.56P A B P A P B = = =
5 5 5 5 6 2. . . .6 6 6 6 1, 2 9
=
7/30/2019 nspch6
36/48
6. Bayes Theorem:
(a)Rule of elimination:
Theorem: If are mutually exclusive eventsof which one must occur, then
(12)
(b)Bayes Theorem: If are mutually exclusive
events of which one must occur, then
(13)
for
1 2, , ...,
nB B B
( )1
( ). ( )n
i i
i
P A P B P A B=
=
1 2, , ..., nB B B
( )
( ) ( )
1
.
( ). ( )
r r
r n
i i
i
P B P A B
P B AP B P A B
=
=
1, 2,......, .r n=
7/30/2019 nspch6
37/48
7/30/2019 nspch6
38/48
Example 25: Using Bayes Theorem
Four technicians regularly make repairs when breakdowns
occur on an automated production line. Jagat, who
services 20% of the breakdowns, makes an incomplete
repair 1 time in 20; Tyagarajan, who services 60% of the
breakdowns, makes an incomplete repair 1 time in 10;
Guna, who services 15% of the breakdowns, makes an
incomplete repair 1 time in 10; and Pandyan, who services
5% of the breakdowns, makes an incomplete repair 1 time
in 20. For the next problem with the production line
diagnosed as being due to an initial repair that wasincomplete, what is the probability that this initial repair
was made by Jagat?
7/30/2019 nspch6
39/48
Solution: Substituting the various probabilities into theformula of Bayes Theorem, we get
and it is of interest to note that although Jagat makes anincomplete repair only 1 out of 20 times, namely, 5% of thebreakdowns, more than 11% of the incomplete repairs arehis responsibility.
7. Mathematical expectation of Decision making.
If the probabilities of obtaining the amounts
are , then themathematical expectation is
(14)
( )1 (0.20)(0.05) 0.114(0.20)(0.05) (0.60)(0.10) (0.15)(0.10) (0.05)(0.05)
P B A = =+ + +
1 2, ,......., ka a or a 1 2, ,......., kp p and p
1 1 2 2 ..... k kE a p a p a p= + + +
7/30/2019 nspch6
40/48
Example 26: Mathematical expectation of net profit
A distributor makes a profit of Rs 200 on an item if it is
shipped from the factory in perfect condition and arriveson time, but it is reduced by Rs 20 if it does not arrive on
time, and by Rs 120 regardless of whether it arrives on
time if it is not shipped from the factory in perfect
condition. If 70% of such items are shipped in perfect
condition and arrive on time, 10% are shipped in perfect
condition but do not arrive on time and 20 % are not
shipped in perfect condition, what is the distributors
expected profit per item?
7/30/2019 nspch6
41/48
Solution: We can argue that the distributor will make the Rs
200 profit 70 % of the time, the Rs 180 profit 10 % of the
time, and the Rs 80 profit 20 % of the time, so that theaverage profit per item ( the expected profit) is
The result is the sum of the products obtained by
multiplying each amount by the corresponding proportion
or probability.
200(0.70) 180(0.10) 80(0.20) 174Rs+ + =
7/30/2019 nspch6
42/48
8.Probability Distributions:
Random variables: They are functions defined over the
elements of a sample space. They are denoted with
capital letters X, Y and so on to distinguish them from their
possible values given in lower case.
The function f(x) = P(X=x) which assigns probability to
each possible outcome x that is called the probabilitydistribution.
Random variables are usually classified according to the
number of values which they can assume. Discrete
random variables can take only a finite number, or acountable infinity of values.
for1( )6
f x = 1,2,3,4,5,6x =
7/30/2019 nspch6
43/48
Gives the probability distribution for the number of pointswe roll with a balanced die.
Since the values of probability distributions are
probabilities and one value of a random variable mustalways occur (i.e)
for all x
and
Example 27: Checking for nonnegativity and totalprobability equals one
Check whether the following can serve as probabilitydistributions:
(a) for x= 1,2,3,4
(b) for x= 0,1,2,3,4
( ) 0f x ( ) 1
all x
f x =
2( )2
xf x =
2
( )
25
xh x =
7/30/2019 nspch6
44/48
Solution: (a) This function cannot serve as a probabilitydistribution because f(1)is negative
(b) The function cannot serve as a probability distribution
because the sum of the five probabilities is andnot 1.
9.Binomial Distribution:
Many statistical problems deal with the situationsreferred to as repeated trials. For example we may wantto know
(i) The probability that one of five rivets will rupture in a
tensile test.
(ii) The probability that 9 of 10 VCRs will run at least1000hours
65
7/30/2019 nspch6
45/48
In determining the probability the following assumptionsare made
1. There are only two possible outcomes for each trial.
2. The probability of a success is the same for each trial.
3. There are ntrials, where nis a constant.
4. The ntrials are independent.
Trials satisfying these assumptions are referred to asBernoulli trials.
Let X: be the random variable that equals thenumber of successes in n trials.
p: Probability of success on any one trial.
(1-p): Probability of failure on any one trial.
The probability of getting x success and (n x) failures in
7/30/2019 nspch6
46/48
The probability of getting x success and (n-x) failures in
some specific order is The number of ways in
which x trials can be selected is , the number of
combinations of xobjects selected from a set of nobjects,and we arrive at
for x= 0,1,2,,n
This probability distributions is called the binomial
distributionbecause for x = 0,1,2,.., and n, the values of
the probabilities are the successive terms of binomial
expansion of ; for the same reason, the
combinatorial quantities are referred to as binomialcoefficients. Actually, the preceding equation defines a
family of probability distributions with each member
characterized by a given value of theparameterp and the
number of trials n.
(1 )x n x
p p
( )nx
( )( ; , ) (1 )x n xnb x n p p px =
[ ](1 )n
p p+
( )nx
7/30/2019 nspch6
47/48
Example 28: Evaluating binomial probabilities
It has been claimed that in 60% of all solar heat installations
the utility bill is reduced by at least one-third.Accordingly, what are the probabilities that the utility bill
will be reduced by at least one-third in
(a) four of five installations;
(b) at least four of five installations?
Solution: (a) Substituting x=4, n=5, and p=0.60 into the
formula for the binomial distribution, we get
( )( )4 5 45(4;5, 0.60) 0.60 (1 0.60) 0.2594b = =
7/30/2019 nspch6
48/48
(b) Substituting x=5, n=5, and p=0.60into the formula for thebinomial distribution, we get
and the answer is
( )( )5 5 55(5;5, 0.60) 0.60 (1 0.60) 0.078
5b = =
(4;5, 0.60) (4;5, 0.60) 0.259 0.078 0.337b b+ = + =