Operations Research topics: SAS/OR,
SAS Optimization, and SAS Simulation Studio

regarding optmization

Occasional Contributor
Posts: 5

regarding optmization

Hi all: I am planning to do the analysis using Response method. I have a Question regarding coding for my independednt variables. For example : X1 , X2, X3 are the independednt coded variables, however in my cases I also have combination tretament (eg. x1+x2). Usullay for the single independednt variables coding can be done as 1, 0, -1. I am very new to the programm so would like to get information on how the coding is done for combination treatment effect? Thank you so much and hope the way I am putting question make sense. 


Example of the scinario: 


Treatment 1: effect of Acid on bacterial growth

Treat 2: Effect of  Protein on bacterial growth 

Treat 3: Effect of  Acid+protein on bacterial growth (How to do coding for combination/interaction effects )





Super User
Posts: 22,818

Re: regarding optmization

Posted in reply to Mralex1234

How does that look like for your data? Are they independent variables or a single variable, treatment?


I might suggest a model as follows:

(Treatment=1)*B1 + (Treatment=2)*B2


So for scenarios:

1. B1

2. B2

3. B1+B2

Occasional Contributor
Posts: 5

Re: regarding optmization

Hi thank you very much for the suggestion.

I have several treatments with different levels and I would like to see the
best combination tretamnets in bacterial growth (response) using
optimization method (eg RSM). I have already done lab experiments and I
have actual results. My aim is to optimize these factors and comapre with
my actual experimental results with predicted results... do they match ?

For example: My independednt variables (I have included only a few here, I
have long list of variables that follows the same trend). I am still
confuse with coding for variables 8 - 15 and a model equation. Looking
forward to get suggestions from you.

All these are independednt variables with different carbs and proteins at
different leveles.

1. control (No treatment)
2. carbohydrate A- 0.1%
3. Carbohydrate B- 0.1%

4. protein A (1%)
5. protein A (2%)
6. Protein B (1%)
7. Protein B (2%)

8. carb A 0.1% + protein A 1%
9. carb A 0.1% + protein A 2%
10. carb A 0.1% + protein B 1%
11. carb A 0.1% + protein B 2%

12. Carb B 0.1% + protein A 1%
13. Carb B 0.1% + protein A 2%
14. Carb B 0.1 % + protein B 1%
15. Carb B 0.1% + Protein B 2%

Once again thank you !!

##- Please type your reply above this line. Simple formatting, no
attachments. -##
Super User
Posts: 22,818

Re: regarding optmization

Posted in reply to Mralex1234

It looks like they're all categorical variables, so you'll need to make indicator variables for these variables. Remember that if you have N levels you need n-1 indicators. Also, you interactions can then be specified as combinations of the other variables.  


Ask a Question
Discussion stats
  • 3 replies
  • 2 in conversation