Cube Plot

Chapter 6
k
The 2 Factorial Design
Chapter 6
Design & Analysis of Experiments

8E 2012 Montgomery
Design of Engineering Experiments

Part 5 The 2k Factorial Design
Text reference, Chapter 6
Special case of the general factorial design; k factors,
all at two levels
The two levels are usually called low and high (they
could be either quantitative or qualitative)
Very widely used in industrial experimentation
Form a basic building block for other very useful
experimental designs (DNA)
Special (short-cut) methods for analysis
We will make use of SAS
Chapter 6

8E 2012 Montgomery
Chapter 6

8E 2012 Montgomery
6.2 The Simplest Case: The 22
- and + denote the low and

high levels of a factor,
respectively
Low and high are arbitrary
terms
Geometrically, the four runs
form the corners of a square
Factors can be quantitative or
qualitative, although their
treatment in the final model
will be different
Chapter 6

8E 2012 Montgomery
Chemical Process Example
A = reactant concentration, B = catalyst amount,

y = recovery
Chapter 6

8E 2012 Montgomery
Analysis Procedure for a

Factorial Design
Estimate factor effects
Formulate model
With replication, use full model
With an unreplicated design, use normal probability
plots
Statistical testing (ANOVA)

Refine the model
Analyze residuals (graphical)
Interpret results
Chapter 6

8E 2012 Montgomery
Estimation of Factor Effects

A y A y A
ab a b (1)
2n
2n
21n [ab a b (1)]
B yB yB
ab b a (1)
2n
2n
21n [ab b a (1)]
See textbook, pg. 235-236 for

manual calculations
The effect estimates are:
A
= 8.33, B = -5.00, AB = 1.67
Practical interpretation?
Design-Expert analysis
ab (1) a b
AB
2n
2n
21n [ab (1) a b]
Chapter 6

8E 2012 Montgomery

Form Tentative Model
Model
Model
Model
Model
Error
Error
Term
Effect
SumSqr
% Contribution
Intercept
A
8.33333
208.333
64.4995
B
-5
75
23.2198
AB
1.66667
8.33333
2.57998
Lack Of Fit 0
0
P Error
31.3333
9.70072
Lenth's ME
Lenth's SME
Chapter 6
6.15809
7.95671

8E 2012 Montgomery
Statistical Testing - ANOVA
The F-test for the model source is testing the significance of the
overall model; that is, is either A, B, or AB or some combination of
these effects important?
Chapter 6

8E 2012 Montgomery

SAS Code and Output
data one;
input A B yield;
cards;
-1 -1 28
-1 -1 25
-1 -1 27
1 -1 36
1 -1 32
1 -1 32
-1 1 18
-1 1 19
-1 1 23
1 1 31
1 1 30
1 1 29
;
proc glm;
class A B;
model yield = A|B;
run;
Chapter 6

8E 2012 Montgomery
10
Chapter 6

8E 2012 Montgomery
11
Residuals and Diagnostic Checking
Chapter 6

8E 2012 Montgomery
12
The Regression Model

Code the levels of factor A and B as follows
Fit regression model
Chapter 6

8E 2012 Montgomery
13

SAS Code and Output
data one;
input A B yield;
cards;
-1 -1 28
-1 -1 25
-1 -1 27
1 -1 36
1 -1 32
1 -1 32
-1 1 18
-1 1 19
-1 1 23
1 1 31
1 1 30
1 1 29
;
proc reg;
model yield = A B;
run;
Chapter 6

8E 2012 Montgomery
14
Source
Model
Error
Corrected Total
Variable
Intercept
A
B
Chapter 6
DF
1
1
1
DF
2
9
11
Analysis of Variance
Sum of
Mean
Squares
Square
F Value
283.33333
141.66667
32.14
39.66667
4.40741
323.00000
Parameter Estimates
Parameter
Standard
Estimate
Error
t Value
27.50000
0.60604
45.38
4.16667
0.60604
6.88
-2.50000
0.60604
-4.13

8E 2012 Montgomery
Pr > F
<.0001
Pr > |t|
<.0001
<.0001
0.0026
15
The Response Surface
Chapter 6

8E 2012 Montgomery
16
6.3 The 23 Factorial Design
Chapter 6

8E 2012 Montgomery
17
Effects in The 23 Factorial Design

A y A y A
B yB yB
C yC yC
etc, etc, ...
Analysis
done via
computer
Chapter 6

8E 2012 Montgomery
18
An Example of a 23 Factorial Design
A = gap, B = Flow, C = Power, y = Etch Rate

Chapter 6

8E 2012 Montgomery
19
Table of and + Signs for the 23 Factorial Design (pg. 244)
Chapter 6

8E 2012 Montgomery
20
Properties of the Table
Except for column I, every column has an equal number of + and

signs
The sum of the product of signs in any two columns is zero
Multiplying any column by I leaves that column unchanged (identity
element)
The product of any two columns yields a column in the table:
A B AB
AB BC AB 2C AC
Orthogonal design
Orthogonality is an important property shared by all factorial designs
Chapter 6

8E 2012 Montgomery
21
Chapter 6

8E 2012 Montgomery
22
ANOVA Summary Full Model
Chapter 6

8E 2012 Montgomery
23
Model Coefficients Full Model
Chapter 6

8E 2012 Montgomery
24
Refine Model Remove Nonsignificant Factors
Chapter 6

8E 2012 Montgomery
25
Model Coefficients Reduced Model
Chapter 6

8E 2012 Montgomery
26
Model Summary Statistics for Reduced Model
R2 and adjusted R2
SS Model 5.106 10
0.9608
R
5
SST
5.314 10
5
2
Adj
SS E / df E
20857.75 /12
1
1
0.9509
5
5.314 10 /15
SST / dfT
R2 for prediction (based on PRESS)

PRESS
37080.44
2
RPred 1
1
0.9302
5
SST
5.314 10
Chapter 6

8E 2012 Montgomery
27
Model Summary Statistics

Standard error of model coefficients (full
model)
se( ) V ( )
2
n2
MS E
2252.56
11.87
k
n2
2(8)
Confidence interval on model coefficients
t / 2,df se( ) t / 2,df se( )

E
Chapter 6

8E 2012 Montgomery
28
Chapter 6

8E 2012 Montgomery
29
Model Interpretation
Cube plots are
often useful visual
displays of
experimental
results
Chapter 6

8E 2012 Montgomery
30
Cube Plot of Ranges
What do the
large ranges
when gap and
power are at the
high level tell
you?
Chapter 6

8E 2012 Montgomery
31
Chapter 6

8E 2012 Montgomery
32
More on the 23 Factorial Design
See p242-244 for definitions of

effects
See p251-252 for Other Methods
for Judging the significance of
Effects.
Chapter 6

8E 2012 Montgomery
33
6.4 The General 2k Factorial Design

Section 6-4, pg. 253, Table 6-9, pg. 254
There will be k main effects, and
k
two-factor interactions
2
k
three-factor interactions
3
1 k factor interaction
Chapter 6

8E 2012 Montgomery
34
6.5 Unreplicated 2k Factorial Designs

These are 2k factorial designs with one
observation at each corner of the cube
An unreplicated 2k factorial design is also
sometimes called a single replicate of the 2k
These designs are very widely used
Risksif there is only one observation at each
corner, is there a chance of unusual response
observations spoiling the results?
Modeling noise?
Chapter 6

8E 2012 Montgomery
35
Spacing of Factor Levels in the

Unreplicated 2k Factorial Designs
If the factors are spaced too closely, it increases the chances

that the noise will overwhelm the signal in the data
More aggressive spacing is usually best
Chapter 6

8E 2012 Montgomery
36
Unreplicated 2k Factorial Designs

Lack of replication causes potential problems in
statistical testing
Replication admits an estimate of pure error (a better
phrase is an internal estimate of error)
With no replication, fitting the full model results in zero
degrees of freedom for error
Potential solutions to this problem

Pooling high-order interactions to estimate error
Normal probability plotting of effects (Daniels, 1959)
Other methodssee text
Chapter 6

8E 2012 Montgomery
37
Example of an Unreplicated 2k Design

A 24 factorial was used to investigate the
effects of four factors on the filtration rate of a
resin
The factors are A = temperature, B = pressure,
C = mole ratio, D= stirring rate
Experiment was performed in a pilot plant
Chapter 6

8E 2012 Montgomery
38
The Resin Plant Experiment
Chapter 6

8E 2012 Montgomery
39
The Resin Plant Experiment
Chapter 6

8E 2012 Montgomery
40
Chapter 6

8E 2012 Montgomery
41
Estimates of the Effects
Chapter 6

8E 2012 Montgomery
42
The Half-Normal Probability Plot of Effects
Chapter 6

8E 2012 Montgomery
43
Design Projection: ANOVA Summary for

the Model as a 23 in Factors A, C, and D
Chapter 6

8E 2012 Montgomery
44
Chapter 6

8E 2012 Montgomery
45
Model Residuals are Satisfactory
Chapter 6

8E 2012 Montgomery
46
Model Interpretation Main Effects

and Interactions
Chapter 6

8E 2012 Montgomery
47
Model Interpretation Response

Surface Plots
With concentration at either the low or high level, high temperature and
high stirring rate results in high filtration rates
Chapter 6

8E 2012 Montgomery
48
Outliers: suppose that cd = 375 (instead of 75)
Chapter 6

8E 2012 Montgomery
49
Chapter 6

8E 2012 Montgomery
50
Dealing with Outliers
Replace with an estimate

Make the highest-order interaction zero
In this case, estimate cd such that ABCD = 0
Analyze only the data you have
Now the design isnt orthogonal
Consequences? Causing significant problems
in interpreting the normal probability plot.
Chapter 6

8E 2012 Montgomery
51
The Drilling Experiment

Example 6.3
A = drill load, B = flow, C = speed, D = type of mud,

y = advance rate of the drill
Chapter 6

8E 2012 Montgomery
52
Normal Probability Plot of Effects

The Drilling Experiment
Chapter 6

8E 2012 Montgomery
53
Residual Plots
R esiduals vs. Pred icted
D ESI GN -EX PER T P lo t

a d v. _ ra te
2.5 8625
Re sid ual s
1.4 4875
0.3 1125
-0.82625
-1.96375
1.69
4.70
7.70
10.71
13.7 1
P red icte d
Chapter 6

8E 2012 Montgomery
54
Residual Plots
The residual plots indicate that there are problems
with the equality of variance assumption
The usual approach to this problem is to employ a
transformation on the response
Power family transformations are widely used
y y
*
Transformations are typically performed to

Stabilize variance
Induce at least approximate normality
Simplify the model
Chapter 6

8E 2012 Montgomery
55
Selecting a Transformation
Empirical selection of lambda
Prior (theoretical) knowledge or experience can
often suggest the form of a transformation
Analytical selection of lambdathe Box-Cox
(1964) method (simultaneously estimates the
model parameters and the transformation
parameter lambda)
Box-Cox method implemented in Design-Expert
Chapter 6

8E 2012 Montgomery
56
(15.1)
Chapter 6

8E 2012 Montgomery
57
The Box-Cox Method

DE S IG N-E X P E RT P l o t
a d v._ ra te
Box-C ox Plot for Power Transforms
A log transformation is
recommended
6.85
Lam bda
Cu rre n t = 1
B e st = -0 .2 3
L o w C.I. = -0 .7 9
Hi g h C.I. = 0 .3 2
The procedure provides a

confidence interval on
the transformation
parameter lambda
5.40
Ln(R es idualSS)
Re co m m e n d tra n sfo rm :
Log
(L a m b d a = 0 )
3.95
If unity is included in the

confidence interval, no
transformation would be
needed
2.50
1.05
-3
-2
-1
Lam bda
Chapter 6

8E 2012 Montgomery
58
Effect Estimates Following the

Log Transformation
Three main effects are
large
No indication of large
interaction effects
What happened to the
interactions?
Chapter 6

8E 2012 Montgomery
59
ANOVA Following the Log Transformation
Chapter 6

8E 2012 Montgomery
60
Following the Log Transformation
Chapter 6

8E 2012 Montgomery
61
The Log Advance Rate Model

Is the log model better?
We would generally prefer a simpler model
in a transformed scale to a more
complicated model in the original metric
What happened to the interactions?
Sometimes transformations provide insight
into the underlying mechanism
Chapter 6

8E 2012 Montgomery
62
6.6 Other Examples of

Unreplicated 2k Designs
The sidewall panel experiment (Example 6.4, pg. 271)
Two factors affect the mean number of defects
A third factor affects variability
Residual plots were useful in identifying the dispersion
effect
The oxidation furnace experiment (Example 6.5, pg.

274)
Replicates versus repeat (or duplicate) observations?
Modeling within-run variability
Chapter 6

8E 2012 Montgomery
63
Example 6.6, Credit Card Marketing, page

278
Using DOX in marketing and marketing
research, a growing application
Analysis is with the JMP screening platform
Chapter 6

8E 2012 Montgomery
64
Other Analysis Methods for

Unreplicated 2k Designs6.5
Lenths method (see text, pg. 262)
Analytical method for testing effects, uses an estimate
of error formed by pooling small contrasts
Some adjustment to the critical values in the original
method can be helpful
Probably most useful as a supplement to the normal
probability plot
Conditional inference charts (pg. 264)
Chapter 6

8E 2012 Montgomery
65
Overview of Lenths method
For an individual contrast, compare to the margin of error
Chapter 6

8E 2012 Montgomery
66
Chapter 6

8E 2012 Montgomery
67
Adjusted multipliers for Lenths method

Suggested because the original method makes too many
type I errors, especially for small designs (few contrasts)
Simulation was used to find these adjusted multipliers

Lenths method is a nice supplement to the normal
probability plot of effects
JMP has an excellent implementation of Lenths method
in the screening platform
Chapter 6

8E 2012 Montgomery
68
Chapter 6

8E 2012 Montgomery
69
6.7 The 2k design and design optimality

The model parameter estimates in a 2k design (and the effect estimates) are
least squares estimates. For example, for a 22 design the model is
y 0 1 x1 2 x2 12 x1 x2
(1) 0 1 (1) 2 (1) 12 (1)(1) 1

a 0 1 (1) 2 (1) 12 (1)(1) 2
b 0 1 (1) 2 (1) 12 (1)(1) 3
ab 0 1 (1) 2 (1) 12 (1)(1) 4
The four
observations
from a 22 design
0
1
1 1 1 1
(1)

1 1 1 1
a
, 1 , 2
y = X + , y , X
2
3
1 1 1 1
b

1 1 1 1
ab
12
4
Chapter 6

8E 2012 Montgomery
70
The least squares estimate of is
= (XX)-1 Xy
The usual contrasts

1
(1) a b ab
a ab b (1)
b ab a (1)
a
b
ab
(1)
(1) a b ab
4
0
(1) a b ab a ab b (1)

1 1 a ab b (1)
4
I 4 b ab a (1) b ab a (1)
2 4
(1) a b ab
12
(1) a b ab
4
0
Chapter 6
0 0 0
4 0 0
0 4 0
0 0 4

8E 2012 Montgomery
The XX matrix is
diagonal
consequences of an
orthogonal design
The regression
coefficient estimates
are exactly half of the
usual effect estimates
71
The matrix XX has interesting and useful properties:
V ( ) 2 (diagonal element of (XX) 1 )
Minimum possible
value for a four-run
design
Maximum possible
value for a four-run
design
|(XX) | 256
Notice that these results depend on both the design that you
have chosen and the model
What about predicting the response?
Chapter 6

8E 2012 Montgomery
72
V [ y ( x1 , x2 )] 2 x(XX)-1 x
x [1, x1 , x2 , x1 x2 ]
V [ y ( x1 , x2 )]
(1 x12 x22 x12 x22 )
4
The maximum prediction variance occurs when x1 1, x2 1
V [ y ( x1 , x2 )] 2
The prediction variance when x1 x2 0 is
V [ y ( x1 , x2 )]
4
What about average prediction variance over the design space?
Chapter 6

8E 2012 Montgomery
73
Average prediction variance

1 1
1
I V [ y ( x1 , x2 )dx1dx2
A 1 1
A = area of design space = 22 4
1 1
1
2 1
(1 x12 x22 x12 x22 )dx1dx2
4 1 1 4
4 2
Chapter 6

8E 2012 Montgomery
74
FDS Graph
Design-Expert Software
Min StdErr Mean: 0.500
Max StdErr Mean: 1.000
Cuboidal
radius = 1
Points = 10000
1.000
StdErr Mean
0.750
0.500
0.250
0.000
0.00
0.25
0.50
0.75
1.00
Fraction of Design Space
Chapter 6

8E 2012 Montgomery
75
For the
2
2 and
in general the
k
2
The design produces regression model coefficients that

have the smallest variances (D-optimal design)
The design results in minimizing the maximum
variance of the predicted response over the design space
(G-optimal design)
The design results in minimizing the average variance
of the predicted response over the design space (Ioptimal design)
Chapter 6

8E 2012 Montgomery
76
Optimal Designs
These results give us some assurance that
these designs are good designs in some
general ways
Factorial designs typically share some (most)
of these properties
There are excellent computer routines for
finding optimal designs (JMP is outstanding)
Chapter 6

8E 2012 Montgomery
77
6.8 Addition of Center Points

to a 2k Designs
Based on the idea of replicating some of the
runs in a factorial design
Runs at the center provide an estimate of
error and allow the experimenter to
distinguish between two possible models:
k
First-order model (interaction) y 0 i xi ij xi x j

i 1
i 1 j i
Second-order model y 0 i xi ij xi x j ii xi2

i 1
Chapter 6
i 1 j i

8E 2012 Montgomery
i 1
78
Chapter 6

8E 2012 Montgomery
79
yF yC no "curvature"
The hypotheses are:
k
H 0 : ii 0
i 1
k
H1 : ii 0
i 1
SS Pure Quad
nF nC ( yF yC ) 2
nF nC
This sum of squares has a

single degree of freedom
Chapter 6

8E 2012 Montgomery
80
Example 6.7, Pg. 286

Refer to the original experiment
shown in Table 6.10. Suppose that
four center points are added to this
experiment, and at the points x1=x2
=x3=x4=0 the four observed
filtration rates were 73, 75, 66, and
69. The average of these four center
points is 70.75, and the average of
the 16 factorial runs is 70.06.
Since are very similar, we suspect
that there is no strong curvature
present.
Chapter 6
nC 4
Usually between 3
and 6 center points
will work well
Design-Expert
provides the analysis,
including the F-test
for pure quadratic
curvature

8E 2012 Montgomery
81
Chapter 6

8E 2012 Montgomery
82
ANOVA for Example 6.7
Chapter 6

8E 2012 Montgomery
83
If curvature is significant, augment the design with axial runs to

create a central composite design. The CCD is a very effective design
for fitting a second-order response surface model
Chapter 6

8E 2012 Montgomery
84
Practical Use of Center Points (pg. 289)

Use current operating conditions as the center
point
Check for abnormal conditions during the
time the experiment was conducted
Check for time trends
Use center points as the first few runs when there
is little or no information available about the
magnitude of error
Center points and qualitative factors?
Chapter 6

8E 2012 Montgomery
85
Center Points and Qualitative Factors
Chapter 6

8E 2012 Montgomery
86
6.9
Why We Work with Coded Design Variables
P 290 -292.
Chapter 6

8E 2012 Montgomery
87

Cube Plot

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Cube Plot

Загружено:

Авторское право:

Доступные форматы

Chapter 6

Design & Analysis of Experiments

Design of Engineering Experiments

Design & Analysis of Experiments

Design & Analysis of Experiments

6.2 The Simplest Case: The 22

- and + denote the low and

Design & Analysis of Experiments

Chemical Process Example

A = reactant concentration, B = catalyst amount,

Design & Analysis of Experiments

Analysis Procedure for a

Statistical testing (ANOVA)

Design & Analysis of Experiments

Estimation of Factor Effects

See textbook, pg. 235-236 for

Design & Analysis of Experiments

Estimation of Factor Effects

Design & Analysis of Experiments

Statistical Testing - ANOVA

Design & Analysis of Experiments

Statistical Testing - ANOVA

Design & Analysis of Experiments

Statistical Testing - ANOVA

Design & Analysis of Experiments

Residuals and Diagnostic Checking

Design & Analysis of Experiments

The Regression Model

Fit regression model

Design & Analysis of Experiments

The Regression Model

Design & Analysis of Experiments

The Regression Model

Design & Analysis of Experiments

The Response Surface

Design & Analysis of Experiments

6.3 The 23 Factorial Design

Design & Analysis of Experiments

Effects in The 23 Factorial Design

Design & Analysis of Experiments

An Example of a 23 Factorial Design

A = gap, B = Flow, C = Power, y = Etch Rate

Design & Analysis of Experiments

Table of and + Signs for the 23 Factorial Design (pg. 244)

Design & Analysis of Experiments

Properties of the Table

Except for column I, every column has an equal number of + and

Design & Analysis of Experiments

Estimation of Factor Effects

Design & Analysis of Experiments

ANOVA Summary Full Model

Design & Analysis of Experiments

Model Coefficients Full Model

Design & Analysis of Experiments

Refine Model Remove Nonsignificant Factors

Design & Analysis of Experiments

Model Coefficients Reduced Model

Design & Analysis of Experiments

Model Summary Statistics for Reduced Model

R2 for prediction (based on PRESS)

Design & Analysis of Experiments

Model Summary Statistics

Confidence interval on model coefficients