How can I implement a multilayer social network in R?

How to implement a simple Bayes Network to support Hiring Decision?

• Hi all, I am thinking of implementing a simple Bayes Network to decide if a candidate should be hired as an exercise. I am thinking the network could look like this Interviewer1    Interviewer2 ....Interviewer n           \                |                 /            \               |                /             \              |               /                   Should HIRE Any suggestion on how should I feed the original data? Should I model the problem differently? Any other thought? Thanks, --Hai

• Answer:

  Actually, I would argue that what you have is one boolean class variable, "DESERVES TO BE HIRED". This has many children, which correspond to the performance of the candidate with each interviewer. This effectively gives you a very simple Naive Bayes classifier. Some justifications: Direction of causality should be that the candidate's ability influences their performance. I think it's not an unreasonable assumption that a candidate's performance with each interviewer is conditionally independent of the his/her other interviews, assuming that questions are chosen at random.