Using Bayes Net Toolbox for Matlab

Question

I want to use Bayes Net Toolbox in matlab , especially score_dags(data, ns, dags) function .

I have:

-3 nodes

-All combinational subset of these nodes that create dag (will be 25 dags)

-Array with size (3,30) where each rows represent the data of nodes (e.g. row 1 for node 1)

When I call the function , if the parameter “data” is the array , and the parameter “dags” is the 25 dags then what it “ns” parameter?

I really appreciate any help you can provide.

Answer 1

The parameter "ns" is the node sizes, where ns(i) is "...the number of values that node i can take on, or the length of node i if i is a continuous-valued vector."

In other words, "ns" holds the domain sizes of each node. For example, if node i can take on values {'red', 'green', 'blue'}, then ns(i) = 3.

It is documented in the .m file for mk_bnet (score_dags passes "ns" to score_family, which then passes it to mk_bnet).