I have shown customers two products with certain attributes and asked them to select one of the products. I have done this 12 times with different combinations for each customer. I have a total number of 1000 customers.

Is there a modeling technique that I can apply to calculate which attributes most customers prefer? If possible, can someone also point me to a working example in R or any other software?


1 Answer 1


What you are looking for is discrete choice modelling.
I would start with a multinomial logit (MNL) model which will model the choice probability as a function of product attributes.
This type of model can easily be implemented in R with many packages:
"mlogit" (https://cran.r-project.org/web/packages/mlogit/vignettes/mlogit.pdf).
You could also use the "clogit" function from the "survival" package (https://stat.ethz.ch/R-manual/R-devel/library/survival/html/clogit.html).
For more sophisticated choice models there are many other packages: gmnl, bayesm, rsghb, etc.

  • $\begingroup$ Thanks for your help. Another quick question as a follow up. In my 12 questions asked to the customers my attributes are changing each time. Is it fine to go ahead with discrete choice modeling $\endgroup$ Commented May 20, 2017 at 15:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.