naive.getug.multinom {lcd}R Documentation

A naive function to get an undirected graph for multinomial data

Description

Learns an undirected independence graph from a given data set. The data are assumed to be multinomially distributed.

Usage

naive.getug.multinom(freq.tb, p.value, method = "mkb")

Arguments

freq.tb an object of class freq.tb, the frequency table for the data.
p.value the thresholding p-value for each conditional independence test.
method a string to specify a method, see details.

Details

Currently, there are three supported method's. "mkb" stands for grow-shrink Markov blanket selection, where greedy Markov blanket selection is performed for each vertex and the blankets are the neighborhood for vertices of the output graph. The "simple" method simply performs conditional independence test for each vertex pair with all the rest vertices as the conditioning set. The "fwd" method uses the forward selection procedure described in Edwards (2000).

Value

The adjacency matrix of an undirected graph.

Author(s)

Zongming Ma and Xiangrui Meng

References

Edwards, D. (2000). Introduction to Graphical Modelling. 2nd Ed. Springer-Verlag, New York.


[Package lcd version 0.7-2 Index]