Professional Documents
Culture Documents
Design of Engineering Experiments Part 5 - The 2 Factorial Design
Design of Engineering Experiments Part 5 - The 2 Factorial Design
Design of Engineering Experiments Part 5 - The 2 Factorial Design
Lenth's ME 6.15809
Lenth's SME 7.95671
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?
A y A y A
B yB y B
C yC yC
etc, etc, ...
Analysis
done via
computer
A B AB
AB BC AB 2C AC
• Orthogonal design
• Orthogonality is an important property shared by all factorial designs
SS E / df E 20857.75 /12
R 2
Adj 1 1 0.9509
SST / dfT 5.314 10 /15
5
What do the
large ranges
when gap and
power are at the
high level tell
you?
With concentration at either the low or high level, high temperature and
high stirring rate results in high filtration rates
Chapter 6 Design & Analysis of Experiments 43
7E 2009 Montgomery
Outliers: suppose that cd = 375 (instead of 75)
1. 44875
R es id uals
0. 31125
- 0. 82625
- 1. 96375
Pre d i c te d
y y
*
• Transformations are typically performed to
– Stabilize variance
– Induce at least approximate normality
– Simplify the model
La mb da
Curre nt = 1
Best = -0 .23
Lo w C. I. = -0 .7 9
High C.I. = 0 .3 2
the transformation
3.95
parameter lambda
If unity is included in the
2.50 confidence interval, no
transformation would be
needed
1.05
-3 -2 -1 0 1 2 3
Lam bda
y 0 1 x1 2 x2 12 x1 x2
(1) 1 1 1 1 0 1
a 1 1 1 1
y = Xβ + ε, y , X ,β 1 ,ε 2
b 1 1 1 1 2 3
ab
1 1 1 1
12 4
Chapter 6 Design & Analysis of Experiments 64
7E 2009 Montgomery
The least squares estimate of β is
(1) a b ab
4
ˆ0 (1) a b ab a ab b (1)
ˆ1 1 a ab b (1) 4
ˆ I 4 b ab a (1) b ab a (1) The regression
2 4 coefficient estimates
ˆ (1) a b ab 4 are exactly half of the
12 (1) a b ab ‘usual” effect estimates
4
Chapter 6 Design & Analysis of Experiments 65
7E 2009 Montgomery
The matrix XX has interesting and useful properties:
Maximum possible
|(XX) | 256 value for a four-run
design
Notice that these results depend on both the design that you
have chosen and the model
What about predicting the response?
FDS Graph
1.000
0.750
0.250
0.000
nF nC ( yF yC ) 2
SS Pure Quad
nF nC