Select Optimal Design Overview
see also  

The purpose of an optimal design is to select design points according to some criteria. Minitab's optimal design capabilities can be used with general full factorial designs, response surface designs, and mixture designs. You can use Select Optimal Design to:

Task

Use to...

Select an optimal design

Select design points from a candidate set to achieve an optimal design. Select optimal design is often used to reduce the number of experimental runs when the original design contains more points than are feasible due to time or financial constraints.

Augment an existing design

Add design points to either D-optimal or distance-based designs. This may be useful if you determine you have additional resources to collect more data after you already generated an optimal design and collected data.  

Improve the D-optimality of a design

Add or exchange points to improve the D-optimality of the design. You can not improve distance-based designs.

Evaluate a design

Obtain optimality statistics for your design. You can use this information to compare designs or to evaluate changes in the optimality of a design if you change the model.

For example, you generate a D-optimal design for a certain model, but then decide to fit the model with different terms. You can determine the change in optimality using the Evaluate design task.

Minitab provides two optimality criteria for the selection of design points:

·    D-optimality - A design selected using this criterion minimizes the variance in the regression coefficients of the fitted model. You specify the model, then Minitab selects design points that satisfy the D-optimal criterion from a set of candidate design points.

·    Distance-based optimality - A design selected using this criterion spreads the design points uniformly over the design space. Use the distance-based method when it is not possible or desirable to select a model in advance. Distance-based optimality is not available with general full factorial designs.