Statistics Toolbox

Controlling Candidate Points

The rowexch function generates a candidate set of possible design points, and then uses a D-optimal algorithm to select a design from those points. It does this by invoking the candgen and candexch functions. If you need to supply your own candidate set, or if you need to modify the one that the candgen function provides, then you may prefer to call these functions separately.

This example creates a design that represents proportions of a mixture, so the sum of the proportions cannot exceed 1.

• % Generate a matrix of (x,y) values with x+y<=1 
  [x,y]=meshgrid(0:.1:1); 
  xy = [x(:) y(:)]; 
  xy = xy(sum(xy,2)<=1,:);
  
  % Compute quadratic model terms for these points. 
  f = x2fx(xy,'q');
  
  % Generate a 10-point design and display it 
  r=candexch(f,10); 
  xy(r,:)
  ans = 
          0           0
          0      1.0000
     1.0000           0
          0      0.5000
     0.5000           0
          0      1.0000
     1.0000           0
     0.5000      0.5000
     0.5000           0
     0.5000      0.5000
  

Designing Experiments with Uncontrolled Inputs

Including Categorical Factors