nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Expanding an existing Custom Design, with additional categorical factors

Jon_Armer

Hello,

I recently performed a custom design with 7x continuous and 1x 2-level categorical factors. Everything went well, and we moved forward with the suggested output.

We're revisiting this work and now need to expand by adding 2 additional catageorical factors into the design and i'm struggling to think of the best way to do it to make the most of the existing data. Fold overs or space filling designs don't seem to let me add additional factors into the grid.

To make it slighlty more complicated, we've also discounted one of the original catagoricals, so are only interested in the one of the original and the two new ones. Everything else can 'exist' in the original design space.

Does anyone have any suggestions?

Thanks,

Jon

Accepted answers

Phil_Kay

Hi Jon,

Good question. Adding categorical factors doesn't seem to be simple.

To add a continuous factor you would typically start with your existing data table and add a column for the additional factor. This column would normally have a constant value for every row because you will not have varied this factor. You can then use DOE > Augment Design and include the additional factor in the X, Factor role along with the existing factors. You can then change the range over which you would like to experiment using that factor.

So a workaround for you would be to specify your additional factors like this first as continuous factors and give them a range of -1 to 1 (or any other 2 values - it doesn't matter) in the Augment Design. Then click Augment. You need to specify main effects for these factors, and whatever model effects for your other factors - don't add any model effects for any factors that are not of interest. When you have generated the design you can then recode the additional factors to have the categorical levels that you will actually use.

Not ideal, I know. Maybe someone else can think of a better way.

Phil

All comments

Phil_Kay

Jon_Armer

Thanks Phil,

I'll give that a try and let you know how it works.

Cheers,

Jon

Jon_Armer

Hi Phil,

In the end I did a mixture of what you suggested and a normal augmentation.

Adding the extra column allowed me to examine an additional continuous factor, and then when expanding the categoricals, I was able to add another level as part of the augmentation itself using the drop-down arrow in the factor dialouge box.

This allowed me to replace the existing factor we had discounted (by simply overwriting) whilst adding the required additional level to the categorical.

I ended up adding additional runs to get the suitable coverage, but even with that increase it was still much better than any solution i'd managed to date (and still used less overall runs). So all in all a great solution.

Cheers,

Jon

Phil_Kay

Great. Thanks for letting us know about your experimental design solution, Jon. I hope you are also successful with the related scientific objectives.