Вы находитесь на странице: 1из 28

BITS Pilani

Pilani Campus

<METHODS IN
SYSTEM
ENGINEERING>
PROF V.MURALIDHAR

MS(SOFTWARE ENGINEERING)
(TECHNIP)-(2013-2014) BATCH

Data Analysis and Quality


Assurance
Control charts for Mean :
Upper control Limit
Central Line
Lower Control Limit
( x , s ) chart : LCL x A1 s
CL

UCL x A1 s
where A1 3 / c2

n is tabulated in tables for

different values of n.
( x , R ) chart : LCL x A2 R
CL

UCL x A2 R
where A2 3 / d 2

n is tabulated in tables for

different values of n.
BITS Pilani, Pilani Campus

Control charts for Range


(R-chart)
S tan dard specified :
LCL D1 '
CL

d 2 '

UCL D2 '
where D1 d 2 3D and D2 d 2 3D and d 2
are obtained from tables.
S tan dard not specified :
LCL D3 R
CL

UCL D4 R
where D3 1 3( D / d 2 ) and
D4 1 3( D / d 2 )
are obtained from tables.

BITS Pilani, Pilani Campus

INTERPRETATION OF MEAN
CHART AND R-CHART
The choice between the mean and R-chart is a managerial
problem. Mean chart is used to control the quality
averages of the samples drawn from a given process,
where as R-Chart is used to control the quality variability's
of the samples.
Also it is better to construct R-Chart first. If it indicates that
the dispersion of the quality by the process is out of
control, it is better not to construct Mean chart, until the
quality dispersion is brought under the control.
Mean chart reveals undesirable variations between samples
while R-Chart reveals any undesirable variations within the
samples
BITS Pilani, Pilani Campus

problems
1. A drilling machine bores holes with a mean diameter of
0.5230cm and a standard deviation of 0.0032cm.
Calculate the 2-sigma and 3-sigma upper and lower
control limits for means of samples of 4 and prepare a
control charts.
[Ans: LCL=0.5198cm,UCL=0.5262cm
LCL=0.5182cm,UCL=0.5278CM]

BITS Pilani, Pilani Campus

Control charts for number


of defectives (or np-Chart)
Suppose random samples of size n are taken from the
product output at some time intervals. If d be the number
of defectives in a sample, then the fraction defective in
that sample is given by p=d/n.
We may suppose that the number of defectives d=np is
binomial with E(np)=nP and V(np)= nP (1 P )
Where P is the population fraction defective.
The process will be in statistical control if P is the same for
all the samples.This will be so if the statistic np lies
within the 3-sigma limits nP3 nP (1 P )

BITS Pilani, Pilani Campus

np-Chart
S tan dard specified :
LCL nP '3
CL

nP ' (1 P ' )

nP '

UCL nP '3

nP ' (1 P ' )

S tan dard not specified :


calculate the mean fraction defective( p )
LCL np 3
CL

np (1 p )

np

UCL np 3

np (1 p )

BITS Pilani, Pilani Campus

Control chart for Fraction


Defective ( or p-chart)
S tan dard specified :
LCL p ' A p ' (1 p ' )
CL

p'

UCL np ' A np ' (1 p ' )


where A 3 / n is tabulated in tables
S tan dard not specified :
calculate the mean fraction defective( p )
LCL p A p (1 p )
CL

UCL p A p (1 p )
where A 3 /

n is tabulated in tables

BITS Pilani, Pilani Campus

Choice between p-chart and


np-chart
When the sample size is constant, it is immaterial whether
one uses np-chart or p-chart. However, when the sample
size varies, all the three control lines will vary with n, in
the case of an np-chart. The resulting chart will be highly
confusing.
On the other hand in a p-chart the central line remains
invariant. Therefore in case the sample size varies, it is
simpler and preferable to use the p-chart.

BITS Pilani, Pilani Campus

Control charts for the number


of defects (or c-chart)
Defective : It is an item that fails to satisfy some given
specification(s)
Defect : It is an event of the items lack of conformity to
given specifications.
Every defectve item contains one or more defects.
Examples : defective rivets in an aircraft , exposed
wires in a refrigerator, surface defects in a roll of a
paper, Crumpled pages in a book, loose screws in a
bicycle etc.

BITS Pilani, Pilani Campus

C-chart control limits


S tan dard specified :
LCL c '3 c '
CL c '
UCL c '3 c '
S tan dard not specified :
calculate the mean number of defects (c )
LCL c 3 c
CL

UCL c 3 c

BITS Pilani, Pilani Campus

Control charts for per unit


Defects ( or u-chart)
The c-chart relates to samples of constant size. In case of
varying sample sizes, it is more convenient to study the
control charts for the number of defects per unit, u=c/n.

BITS Pilani, Pilani Campus

Control limits for u-chart

S tan dard specified :


LCL u '3 u ' / n
CL u '
UCL u '3 u ' / n where u is the mean per unit defects
S tan dard not specified :
calculate the mean number of defects(c)
LCL u 3 u / n
CL u
UCL u 3 u / n

BITS Pilani, Pilani Campus

Applications of u-chart
An u-chart is applied to control the number of defects per
unit in the case of fairly complex assembled units as
T.V.sets, aircraft engines, refregirators etc.

BITS Pilani, Pilani Campus

FORECASTING AND
REGRESSION MODELS
In Decision making we deal with devising Future Plans :
For instance, in Financial Planning like share markets,
Investments in Fixed and Recurring deposits, we need to
predict the pattern of cash-flow over time.

BITS Pilani, Pilani Campus

THREE TECHNIQUES FOR


FORECASTING FUTURE VARIATIONS
Moving Average Technique
Exponential Smoothening
Regression

BITS Pilani, Pilani Campus

Moving Average Technique


The value for a period of time t is given by
yt b t

Where the time series is stable, follows a constant process


b = Unknown constant parameter estimated from historical
data

= Random component for period t with zero mean and


constant variance
The data for different periods are not correlated

BITS Pilani, Pilani Campus

Regression method
problems

BITS Pilani, Pilani Campus

Exponential Smoothening
problems

BITS Pilani, Pilani Campus

DECISION THEORY(Sequential
Decision Models)

A Constructive process for


selecting the best alternatives.
The goodness of a selected
alternative depends on the
quality of the data used in
describing the decision
situation
BITS Pilani, Pilani Campus

THREE CATEGORIES OF
DECISION MAKING
I Decision making under certainty in which the data are
known deterministically (Data well defined)
II Decision making under risk in which the data can be
described by probability distribution (Data are
ambiguous)
III Decision making under uncertainty in which the data
cannot be assigned weights that represent their degree
of relevance in the decision process. (Data middle-ofthe-road situation)

BITS Pilani, Pilani Campus

Decision making under certainty in which the


data are known deterministically

Analytic Hierarchy process(AHP) :


Designed for situation in which ideas , feelings and
emotions are quantified based on subjective
judgment to provide a numeric scale for prioritizing
decision alternatives

BITS Pilani, Pilani Campus

problems
1.Mr X has received full academic scholarships from three
Institutions : U of A , U of B and U of C. To select a
university Mr X specifies two main criteria : location and
academic reputation. Being excellent student he judges
academic reputation to be five times as important as
location, which gives a weight of approximately 17% to
location and 83% to reputation. Analyze the best choice
of the university by ranking the three universities from
the standpoint of location and reputation.

BITS Pilani, Pilani Campus

Given the table of ranks based on two


criteria for the three universities :

Two criteria
Calculation

Percent

Weight
estimates

for

Three

Decision

Alternatives

criterion

U of A

U of B

U of C

Location

12.9

22.7

59.4

Reputation

54.5

27.3

18.2

Composite
weight

0.129x0.17
+
0.545x0.83
=

0.17x0.227
+
0.83x0.273
=

0.594x0.17
+
0.182x0.83
=

0.4743

0.2737

0.2520

BITS Pilani, Pilani Campus

Summary of AHP calculations :


DECISION : SELECT A UNIVERSITY
H1

H2
Alt
ern
ati
ves

P1

U
of
A
P1
1

U
of
B
P1
2

P2

U
of
C
P1
3

U
of
A
p2
1

U
of
B

U
of
C
p2
3

P2
2

BITS Pilani, Pilani Campus

Summary of AHP calculations :


DECISION : SELECT A UNIVERSITY
H1

H2
Alt
ern
ati
ves

Q1

U
of
A
Q1
1

U
of
B
Q1
2

Q2

U
of
C
Q1
3

U
of
A
Q2
1

U
of
B

U
of
C
Q2
3

Q2
2

BITS Pilani, Pilani Campus

Determination of relative weights :


nxn comparison matrix
It reflects the decision makers judgment of the relative importance
of the different criteria.
Aij = criteria in the row i is ranked relative to each of the criteria
represented by the n columns.
Also Aij = 1 signifies that i and j are equally important
Aij = 5 indicates that i is strongly more important than j
Aij = 9 indicates that I is extremely more important than j
Other intermediary values between 1 and 9 are interpreted
correspondingly
For consistency Aij = 1 automatically implies that Aij=1/k
Also all the diagonal elements Aii of A must equal 1 because they
rank a criteria against itself.

BITS Pilani, Pilani Campus

Consider the previous


problem
Let L represent location and R represent reputation.
Then the correspnding matrix is given by
L
R
A=L(1
1/5 )
R(5
1)
To determine relative weights :
Normalized matrix is given by
L
R
Row Average
N = L(.17 .17)
WL = .17
R(.83 .83)
WR= .83
The columns are identical implies the charecteristic that occurs only when decision
maker exhibits perfect consistency in specifying the entries of the comparison
matrix A.

BITS Pilani, Pilani Campus

Вам также может понравиться