I'm trying to build a bayesian network for satisfcation survey data. My data is made of 13 questions about services, products etc... each customer can answer from 1 (Very unsatisfied) to 4 (very satisfied) with no "neutral feeling"). There is an other question about the overall satisfaction where the customer can answer 1 (very unsatisfied), 2 (unsatisfied), 3 (satisfied), 4(very satisfied) and 5 (totally satisfied). I plan to use bnlearn to build the bayesian network. The goal is to identify important features who lead customers to give 5 on overall satisfaction item. In your opinion, what is the best way to recode the overall satisfaction : 0 for 1, 2, 3, 4 and 1 for 5 (totally satisfied)? I tried but results let me skeptical.
Thanks for your help.