Math 2.0

Statistics Definition

Fractional factorial design

Builds on Factorial experimentFinite fieldSubgroup

Definition

Let k>p≥1k > p \ge 1, and code the levels of kk two-level factors as ±1\pm 1. Then the runs of the full 2k2^k factorial experiment are the points of {−1,1}k\set{-1, 1}^k. For a subset w⊆{1,…,k}w \subseteq \set{1, \dots, k}, called a word, write xw=∏j∈wxjx^w = \prod_{j \in w} x_j for x∈{−1,1}kx \in \set{-1, 1}^k. Identify words with vectors of 𝔽2k\F_2^k, with the symmetric difference as addition. A regular 2k−p2^{k-p} fractional factorial design is a set of runs F={x∈{−1,1}k:xw=ϵw for all w∈H},F = \set{x \in \set{-1, 1}^k : x^{w} = \epsilon_w \text{ for all } w \in H}, where:

  1. HH is a subgroup of 𝔽2k\F_2^k of order 2p2^p, the defining relation, with no word of size 11;
  2. the signs ϵw∈{−1,1}\epsilon_w \in \set{-1, 1} satisfy ϵ∅=1\epsilon_{\emptyset} = 1 and ϵw△w′=ϵwϵw′\epsilon_{w \triangle w'} = \epsilon_w \epsilon_{w'}, so that FF is not empty.

Then FF has 2k−p2^{k-p} runs. Two effects with words uu and vv are aliased if u△v∈Hu \triangle v \in H. The resolution of the design is the smallest size of a nonempty word of HH.

Remarks

On FF, the contrast columns xux^u and xvx^v of aliased effects agree up to the sign ϵu△v\epsilon_{u \triangle v}, so their effects cannot be separated. Here 𝔽2\F_2 is the field with two elements.

Example: k=3k = 3, p=1p = 1, H={∅,{1,2,3}}H = \set{\emptyset, \set{1,2,3}} with ϵ=+1\epsilon = +1, that is, the generator C=ABC = AB and the defining relation I=ABCI = ABC. The runs are (−1,−1,+1)(-1,-1,+1), (+1,−1,−1)(+1,-1,-1), (−1,+1,−1)(-1,+1,-1), (+1,+1,+1)(+1,+1,+1), four of the eight. The main effect AA, with word {1}\set{1}, is aliased with BCBC, since {1}△{1,2,3}={2,3}\set{1} \triangle \set{1,2,3} = \set{2,3}. The design has resolution III.

Non-example: the four runs with x1=−1x_1 = -1 form the set {x:x{1}=−1}\set{x : x^{\set{1}} = -1}, which uses the word {1}\set{1} of size 11, so clause 1 fails. In these runs factor AA never changes, so its main effect cannot be estimated at all.

Reading path

To understand this page from the ground up, read these 4 pages in order. The foundations come first.

Show the path as a graph
Each page uses the pages above it. The colour of the bar on a box shows how strong the evidence is, and the box also gives it as text. The list has the same pages.
  • no evidence level
  1. Factorial experiment concept
  2. Finite field concept
  3. Subgroup concept
  4. Fractional factorial design definition

Further reading

  1. Web https://en.wikipedia.org/wiki/Fractional_factorial_design
About this page
Tag10WTA1QZXD
URIm2://10WTA1QZXD
Kinddefinition
WritingW0 stub
MSC62K15
DefinesFractional factorial design
UsesFactorial experiment, Finite field, Subgroup

Markdown · JSON · Edit on GitHub

What this page needs for W1

  • an inventory row: the row of the inventory that the node accounts for (0 of 1)
  • a source with a locator (0 of 1)

The writing levels say how complete the text of a page is. They do not measure its truth.

What links here

No page links to this page yet.