learn.tbn: Learn a Tweedie Bayesian Network.
In km20/tbn: Tweedie Bayesian Network

Description Usage Arguments Value Examples

This function builds a Tweedie Bayesian network based on a data set. It combines both structure and parameters learning. If a known graph is provided, only the parameters will be estimated relying on the provided network structure.

1	learn.tbn(data, pgrids, G = NULL, nodes.order = NULL, conf.level = 0.05)

`data`	a data frame containing the sample used to learn the model.
`pgrids`	a list of numeric vectors containing the values among which the power parameters are chosen in order to maximize the model's log-likelihood. If the power parameter is knonw, each pgrid vector will contain a single value.
`G`	(optional) the Tweedie Bayesian network structure as an adjacency matrix i.e., G[i,j] =1 if Xj is a parent of Xi. If no graph is provided, the structure will be learned relying on student statistical tests.
`nodes.order`	(optional) The nodes ordering defining an ancestral ordering of the Bayesian network.
`conf.level`	(optioal) The confidence level for the structure learning. Default value is 0.05.

This function returns a named list consting of the Tweedie bayesian network structure and parameters.

# 3 nodes: the third is node has a Gaussian conditional distribution.
## Not run: 
learn.tbn(data, pgrids=list(c(2,3), c(2.2,2.3,2.4), c(0)))

## End(Not run)