PNetica-package: Parameterized Bayesian Networks Netica Interface

In ralmond/PNetica: Parameterized Bayesian Networks Netica Interface

Parameterized Bayesian Networks Netica Interface

Description

This package provides RNetica implementation of Peanut interface.This provides an implementation of the Peanut protocol using the Netica (RNetica) Bayesian network engine. This allows parameters of parametric Bayesian network models written and Netica and using Peanut to be processed with R tools, and the parameter saved in the Netica objects.

Details

The DESCRIPTION file: This package was not yet installed at build time.

The Peanut package provides a set of generic functions for manipulation parameterized networks, in particular, for the abstract Pnet and Pnode classes. This package provides concrete implementations of those classes using the built in classes of RNetica. In particular, Pnet.NeticaBN extends NeticaBN and Pnode.NeticaNode extends NeticaNode. The documentation object Pnode.States documents additional fields of this object.

The properties of the Pnet and Pnode objects are stored as serialized Netica user fields (see NetworkUserObj and NodeUserObj). The documentation object Pnode.Properties documents the methods.

The as.Pnet (as.Pnode) method for a NeticaBN (NeticaNode) merely adds “Pnet” (“Pnode”) to class(net) (class(node)). All of the methods in the PNetica are defined for either the NeticaBN or NeticaNode object, so strictly speaking, adding the “Pnet” or “Pnode” class is not necessary, but it is recommended in case this is used in the future.

PNetica Specific Implementation Details

Here are some Netica specific details which may not be apparent from the description of the generic functions in the Peanut package.

1. The cases argument to calcPnetLLike, calcExpTables and GEMfit all expect the pathname of a Netica case file (see write.CaseFile).

2. The methods calcPnetLLike, calcExpTables, and therefore GEMfit when called with a Pnet as the first argument, expect that there exists a node set (see NetworkNodesInSet) called “onodes” corresponding to the observable variables in the case file cases.

3. The function CompileNetwork needs to be called before calls to calcPnetLLike, calcExpTables and GEMfit.

4. The method PnetPnodes stores its value in a nodeset called “pnodes”. It is recommended that the accessor function be used for modifying this field.

5. The PnetPriorWeight field of the Pnet.NeticaBN object and all of the fields of the Pnode.NeticaNode are stored in serialized user fields with somewhat obvious names (see NetworkUserObj and NodeUserObj). These fields should not be used for other purposes.

Creating and Restoring Pnet.NeticaBN objects

As both the nodesets and and user fields are serialized when Netica serializes a network (WriteNetworks) the fields of the Pnet.NeticaBN and Pnode.NeticaNode objects should be properly saved and restored.

The first time the network and nodes are created, it is recommended that Pnet and Pnode.NeticaNode (or simply the generic functions Pnet and Pnode. Note that calling Pnode will calculate defaults for the PnodeLnAlphas and PnodeBetas based on the current value of NodeParents(node), so this should be set before calling this function. (See examples).

Index

Index: This package was not yet installed at build time.

Legal Stuff

Netica and Norsys are registered trademarks of Norsys, LLC (http://www.norsys.com/), used by permission.

Extensive use of PNetica will require a Netica API license from Norsys. This is basically a requirement of the RNetica package, and details are described more fully there. Without a license, RNetica and PNetica will work in a student/demonstration mode which limits the size of the network.

Although Norsys is generally supportive of the RNetica project, it does not officially support RNetica, and all questions should be sent to the package maintainers.

Author(s)

Russell Almond

Maintainer: Russell Almond <[e-mail]>

References

Almond, R. G. (2015) An IRT-based Parameterization for Conditional Probability Tables. Paper presented at the 2015 Bayesian Application Workshop at the Uncertainty in Artificial Intelligence Conference.

See Also

PNetica depends on the following other packages.

RNetica

A binding of the Netica C API into R.

Peanut

An the generic functions for which this package provides implementations.

CPTtools

A collection of implementation independent Bayes net utilities.

Examples

sess <- NeticaSession()
startSession(sess)

## Building CPTs
tNet <- CreateNetwork("TestNet", session=sess)


theta1 <- NewDiscreteNode(tNet,"theta1",
                         c("VH","High","Mid","Low","VL"))
NodeLevels(theta1) <- effectiveThetas(NodeNumStates(theta1))
NodeProbs(theta1) <- rep(1/NodeNumStates(theta1),NodeNumStates(theta1))
theta2 <- NewDiscreteNode(tNet,"theta2",
                         c("VH","High","Mid","Low","VL"))
NodeLevels(theta2) <- effectiveThetas(NodeNumStates(theta2))
NodeProbs(theta2) <- rep(1/NodeNumStates(theta2),NodeNumStates(theta2))

partial3 <- NewDiscreteNode(tNet,"partial3",
                            c("FullCredit","PartialCredit","NoCredit"))
NodeParents(partial3) <- list(theta1,theta2)

partial3 <- Pnode(partial3,Q=TRUE, link="partialCredit")
PnodePriorWeight(partial3) <- 10
BuildTable(partial3)

## Set up so that first skill only needed for first transition, second
## skill for second transition; adjust alphas to match
PnodeQ(partial3) <- matrix(c(TRUE,TRUE,
                             TRUE,FALSE), 2,2, byrow=TRUE)
PnodeLnAlphas(partial3) <- list(FullCredit=c(-.25,.25),
                                PartialCredit=0)
BuildTable(partial3)
partial4 <- NewDiscreteNode(tNet,"partial4",
                            c("Score4","Score3","Score2","Score1"))
NodeParents(partial4) <- list(theta1,theta2)
partial4 <- Pnode(partial4, link="partialCredit")
PnodePriorWeight(partial4) <- 10

## Skill 1 used for first transition, Skill 2 used for second
## transition, both skills used for the 3rd.

PnodeQ(partial4) <- matrix(c(TRUE,TRUE,
                             FALSE,TRUE,
                             TRUE,FALSE), 3,2, byrow=TRUE)
PnodeLnAlphas(partial4) <- list(Score4=c(.25,.25),
                                Score3=0,
                                Score2=-.25)
BuildTable(partial4)

## Fitting Model to data

irt10.base <- ReadNetworks(system.file("testnets","IRT10.2PL.base.dne",
                                       package="PNetica"), session=sess)
irt10.base <- as.Pnet(irt10.base)  ## Flag as Pnet, fields already set.
irt10.theta <- NetworkFindNode(irt10.base,"theta")
irt10.items <- PnetPnodes(irt10.base)
## Flag items as Pnodes
for (i in 1:length(irt10.items)) {
  irt10.items[[i]] <- as.Pnode(irt10.items[[i]])
  
}


casepath <- system.file("testdat","IRT10.2PL.200.items.cas",
                        package="PNetica")
## Record which nodes in the casefile we should pay attention to
NetworkNodesInSet(irt10.base,"onodes") <-
   NetworkNodesInSet(irt10.base,"observables")


BuildAllTables(irt10.base)
CompileNetwork(irt10.base) ## Netica requirement

item1 <- irt10.items[[1]]
priB <- PnodeBetas(item1)
priA <- PnodeAlphas(item1)
priCPT <- NodeProbs(item1)

gemout <- GEMfit(irt10.base,casepath)


DeleteNetwork(irt10.base)
DeleteNetwork(tNet)
stopSession(sess)

ralmond/PNetica documentation built on Sept. 19, 2023, 8:27 a.m.