ParDAG-class function

Class "ParDAG" of Parametric Causal Models

This virtual base class represents a parametric causal model.

1.1

class

Details

The class "ParDAG" serves as a basis for simulating observational and/or interventional data from causal models as well as for parameter estimation (maximum-likelihood estimation) for a given causal model in the presence of a data set with jointly observational and interventional data.

The virtual base class "ParDAG" provides a skeleton for all functions relied to the aforementioned task. In practical cases, a user may always choose an appropriate class derived from ParDAG which represents a specific parametric model class. The base class itself does not represent such a model class.

Constructor

new("ParDAG", nodes, in.edges, params)

• nodes: Vector of node names; cf. also field .nodes.
• in.edges: A list of length p consisting of index vectors indicating the edges pointing into the nodes of the DAG.
• params: A list of length p consisting of parameter vectors modeling the conditional distribution of a node given its parents; cf. also field .params.

Fields

• .nodes:: Vector of node names; defaults to as.character(1:p), where p denotes the number of nodes (variables) of the model.
• .in.edges:: A list of length p consisting of index vectors indicating the edges pointing into the nodes of the DAG.
• .params:: A list of length p consisting of parameter vectors modeling the conditional distribution of a node given its parents. The entries of the parameter vectors only get a concrete meaning in derived classes belonging to specific parametric model classes.

Class-Based Methods

• node.count():: Yields the number of nodes (variables) of the model.

• simulate(n, target, int.level):: Generates $n$

   (observational or interventional) samples from the parametric causal model. The intervention target to be used is specified by the parameter `target`; if the target is empty (`target = integer(0)`), observational samples are generated. `int.level` indicates the values of the intervened variables; if it is a vector of the same length as `target`, all samples are drawn from the same intervention levels; if it is a matrix with $n$ rows and as many columns as `target` has entries, its rows are interpreted as individual intervention levels for each sample.
  

• edge.count():: Yields the number of edges (arrows) in the DAG.

• mle.fit(score):: Fits the parameters using an appropriate Score object.

Methods

• plot: signature(x = "ParDAG", y = "ANY"): plots the underlying DAG of the causal model. Parameters are not visualized.

Author(s)

Alain Hauser (alain.hauser@bfh.ch )

See Also

GaussParDAG