Using DoEgen without Y truth

[jcjlin]

Hello DoEgen Team,

First of all, I would like to express my gratitude for your efforts in developing this useful package.
I have a question regarding the analysis capabilities of DoEgen when the Y truth values are not available.
Is it possible to use this package for analysis in such scenarios? If so, could you provide some guidance or reference on how to proceed without the Y truth data? Thanks.

[Daniel-Haas-B]

Hi. I am not from DoEgen team. I have only recently started learning about design of experiments and had the same question. Here is what I understand (I might be wrong!).

This project is based on the oapackage, which uses orthogonal arrays (hence the name) to generate a design of experiment.
In this case, we primarily focus on structuring the experiments by selecting combinations of factor levels, in a way which is statistically relevant. This phase is about planning how to conduct experiments rather than analyzing what the outcomes will be. So from my understanding, the initial design is all done without any "Y truth data", using your words

[GeoMattB]

While the analysis code in DoEval does rely on a "truth" this is primarily because we were using DoEgen to compare models against observations. Have a look at Goos & Jones 2011, you'll see that there are many ways to analyse the results.
If you need to, you can simply use a "dummy" truth value to compare the results. If you enter a 0 or 1 as the truth then the results are essentially compared against their distance to that.