- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Assessment of an A-optimal design
I have created an A-optimal design that is intended to serve as a screening design with the following efficiency values. How would you rate these values? Would this be suitable as a screening design?
$D | |
[1] | 0,8632544 |
$A | |
[1] | 1,208548 |
$Ge | |
[1] | 0,768 |
$Dea | |
[1] | 0,74 |
This post originally written in German and has been translated for your convenience. When you reply, it will also be translated back to German.
Accepted Solutions
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Einschätzung eines A-optimalen Designs
Hi @Thommy7571,
Efficiency values should not be interpreted on their own. See Design Diagnostics
You can try to create several designs, with different optimality criterion, run size, etc... and use the platform Compare Designs to compare and select the design best suited for your use case.
Design efficiencies are only a "summary" metric to help compare the performances of several designs, but I would recommend spending time on the different parts of the Compare Designs platform : Power Analysis, , Prediction Variance Profile , Fraction of Design Space Plot, Relative Estimation Efficiency, Alias Matrix Summary, Absolute Correlations ...
Hope this answer will help you,
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Report Inappropriate Content
Re: Einschätzung eines A-optimalen Designs
Hi @Thommy7571,
Efficiency values should not be interpreted on their own. See Design Diagnostics
You can try to create several designs, with different optimality criterion, run size, etc... and use the platform Compare Designs to compare and select the design best suited for your use case.
Design efficiencies are only a "summary" metric to help compare the performances of several designs, but I would recommend spending time on the different parts of the Compare Designs platform : Power Analysis, , Prediction Variance Profile , Fraction of Design Space Plot, Relative Estimation Efficiency, Alias Matrix Summary, Absolute Correlations ...
Hope this answer will help you,
"It is not unusual for a well-designed experiment to analyze itself" (Box, Hunter and Hunter)