Chapter Contents |
Previous |
Next |

Macros for the Design and Analysis of Experiments |

where *expression* is a valid comparison expression
involving one or more of the following design parameters:

NTMTS | the number of design factors, where . |

NRUNS | the number of experimental runs, where . |

NBLKS | the number of blocks in the design. |

K | the size of each block. |

RES | the resolution of the design, where
RES. |

Each part of *expression* is a comparison involving one or more
of the above design parameters. Each part of *expression* must
be enclosed
in parentheses, and the entire expression must also be enclosed in
parentheses. For example, a simple *expression* that requests
designs with 5 or more treatments is shown below:

%adxgen %adxff %adxpff((ntmts ge 5))

You can use AND and OR operators to form a complex *expression*.
The symbols & and | can also be used for AND and OR, respectively.
For example, the following statements list all regular fractions for
six factors with either less than 80 runs or exactly 8 blocks:

%adxgen %adxff %adxpff((ntmts=6) & ((nruns < 80) or (nblks=8)))

The ADXPFF macro is useful when you do not know the exact parameters of the
design you want to construct. The ADXPFF macro also lists the valid
ADXFFD macro call for each design. Once you have chosen a design from
the list, you can simply submit the associated ADXFFD macro call, filling
in a SAS data set name for the value **data-set-name** shown in the
listing.

For example, suppose you want to find a design

- for 5 or more factors
- in 32 or fewer runs
- of resolution 4 or higher

%adxgen %adxff %adxpff((ntmts >= 5) & (nruns <= 32) & (res >= 4))returns 51 designs, with as many as 11 factors and with blocks as small as 2 runs per block. You can choose one of these 51 designs and then submit the appropriate ADXFFD call to create the design and output it to a data set.

Chapter Contents |
Previous |
Next |
Top |

Copyright © 1999 by SAS Institute Inc., Cary, NC, USA. All rights reserved.