stats cont Flashcards
1
Q
y equation
A
y = mu + tau + beta + epsilon tau = treatment beta = experimental
2
Q
replicate vs repeat
A
repeat - use mean
replicate - error df will be large
3
Q
S^2
A
sum(yi-ybar)^2/(n-1)
4
Q
R^2
A
How much of model can we explain?
5
Q
PRESS
A
pull out data point, predict value, compare to real data pt
6
Q
How to put df in error term
A
- Replicate
- Pool
- Center point
7
Q
Fractionalization element
A
exponent in 2^k factorial
8
Q
Fraction
A
The exponent subtracted in 2^k factorial (it’s 2^the power)
9
Q
Benefit of replicating (higher resolution)
A
- more proof
- better estimate of residual error
10
Q
Benefit of not replicating (lower resolution)
A
- Rely on pooling - more that can probably pool
11
Q
Interaction
A
The effect of one variable is modified by another variable
12
Q
Factorial vs one-at-a-time
A
- OAAT requires more runs (not efficient)
- cannot estimate interactions
- can miss optimal settings of factors
13
Q
RSM
A
- factor screening
- move from initial guess to region of optimum
