How to have subfactors - easy procedure?

[#1]

Hello all,

I want to perform design of experiments and learn maximum about interaction between 3 sub categories.

I want to have model where i have 4 factors - F1, F2, F3 & F4

Under Factor F1, i have 4 subfactors - F1a, F1b, F1c, F1d

Under Factor F2, i have 3 subfactors - F2a, F2b, F2c

Under Factor F3, i have 3 subfactors - F3a, F3b, F3c

and F4 dont have any subfactor. it is just F4.

So, in total, i have 11 factors, but in a single experimental run/treatment i want to have only F1, F2, F3 and F4 tested with various subfactors on 2 levels.

How it can be done, without having any coding skills?

[#2]

Hi @Darshan,

Welcome in the Community !
Can you provide an example (not related to your topic necessarily) about the factors and subfactors ?
I'm not sure if what you're describing is a typical situation with factors and corresponding levels (for example, F1 is the choice of chemical and the levels/choices are F1a, F1b, F1c and F1d), or a situation with possible restriction randomization or blocking (like F1 factors can not change independantly of F2 or other factor family).

[#3]

Hi Victor,

Thank you very much for your reply and warm welcome,

okay example could be, i want to test interaction between Carbon, nitrogen and hydrogen sources with minerals

so, F1 should be all Carbon sources, like Sugars, carbohydrates, polysaccharides

F2 should be Nitrogen sources like proteins, amino acids

F3 should be mineral salts like potassium salt, sodium salt,

And i want to understand how a perticular sugar with perticular protein along with specific salt gives best interaction. in this case, i dont want to identify sugar and polysaccharide interaction, i dont want salt -salt interaction,

so, even though i have more than 10 factors to test, i will put only only 1 subfactor from each category and use only 3 or 4 main factors for testing in each treatment or experimental run. I will not put multiple carbon sources in a single treatment. I am trying to my best to explain, but if you need further clarification, please ask again.

thanks

[#4 · Accepted solution]

Hi @Darshan,

Ok, from what I understand, it seems that F1, F2 and F3 are categorical factors with different levels.
For example,

• F1 is a categorical factor "Carbon", with 3 possible levels to choose from : Sugar / Carbohydrate / Polysaccharide
• F2 is a categorical factor "Nitrogen", with 2 posssible levels to choose from : Protein / Amino-acid
• F3 is a categorical factor "Mineral salts" with 2 posssible levels to choose from : Potassium / Sodium

And besides these factors, you could add continuous factors for the amount of each category like F1_amount, F2_amount, F3_amount and F4_amount (I have added in my example through coded values like -1 and +1, but feel free to use real values for your experiments).

You could then use the Custom Designs platform to set up your factors and define a model according to your requirements/needs, for example with main effects for your factors and 2-factors interactions between your factors.

As an illustration, here is an example with the configuration proposed (and factors for the different categories amounts):

• Defining the factors and model (with main effects and 2-factors interactions):

  

• Generated design with 42 runs:
  

This is just an example to show you the method and design creation that could match your requirement of having only one type of C/N/salts used for each experiment, but still analyzing the interactions between specific choices combination. Maybe you won't need the numeric factors linked to the amount of each category levels. There are of course other methods available and the settings of the design could also be made so that the total amount of "sources" is fixed to 100%, or specific combinations are not used thanks to Define Factor Constraints ... If you need more help or guidance, you can explain what are the other requirements in more details.

Please find attached the datatable with the generated design.

Hope this answer will help you,

[#5]

Million Thanks Victor!!

This is exactly what i was looking for.

Have nice time!!