R language Access Menu

Score-based Structure Learning

Package bnlearn provides functions for constraint-based structure learning and other algorithms to produce directed acyclic graph (DAG) of data. In following examples, the mtcars dataset that has information about features of different cars is used: 

library(bnlearn)
res = gs(mydf)  # Grow-Shrink algorithm
plot(res, main="Simple Learning Bayesian Algorithm")

                      

Other algorithms available include Incremental Association (IAMB), the Fast Incremental Association (Fast IAMB) and the Interleaved Incremental Association (Inter IAMB) constraint-based algorithms.

Structure network using hill-climbing and Tabu greedy search algorithms are also possible: 

Code:

> library(bnlearn)
> res = hc(mtcars)
> plot(res)

                    

This algorithm shows more relationships. 


    Comments & Feedback