generate_mvn_data: Generate data from a DAG

In ccdrAlgorithm: CCDr Algorithm for Learning Sparse Gaussian Bayesian Networks

Generate data from a DAG

Description

Given a Gaussian DAG, generate data from the underlying distribution. Equivalently, generate data from a multivariate normal distribution given one of its SEM. Can generate both observational and intervention data.

Usage

generate_mvn_data(graph, params, n = 1, ivn = NULL, ivn.rand = TRUE)

Arguments

graph	DAG in edgeList format.
params	Vector of parameters. Last p elements correspond to variances (p = number of nodes in graph), initial elements correspond to edge weights.
n	Number of samples to draw.
ivn	List of interventions (see sparsebnData). Must be a list with exactly n components.
ivn.rand	If TRUE, random N(0,1) values will be drawn for each intervention. Otherwise, these values need to supplied manually in ivn.

Details

If ivn = NULL, then n observational samples are drawn. For each component of ivn that is not NULL, interventional samples will be drawn with the values of each node specified in the component.

Examples

### Generate observational data
gr <- sparsebnUtils::random.graph(5, 5) # use sparsebnUtils package to generate a random graph
gr.params <- runif(10) # there are 5 coefficients + 5 variances
data.obs <- ccdrAlgorithm::generate_mvn_data(graph = gr,
                                             n = 100,
                                             params = gr.params)

### Generate experimental data
ivn <- as.list(c(rep("V1", 50), rep("V2", 50))) # 50 interventions on V1, 50 interventions on V2
data.ivn <- ccdrAlgorithm::generate_mvn_data(graph = gr,
                                             n = 100,
                                             params = gr.params,
                                             ivn = ivn)

### Use pre-specified values for interventions
###  In this toy example, we assume that all intervened nodes were fixed to
###  to the value 1, although this can be any number of course.
ivn.vals <- lapply(ivn, function(x) sapply(x, function(x) 1)) # replace all entries with a 1
data.ivn <- ccdrAlgorithm::generate_mvn_data(graph = gr,
                                             n = 100,
                                             params = gr.params,
                                             ivn = ivn.vals,
                                             ivn.rand = FALSE)

### If ivn.rand = FALSE, you must specify values
###  The code below will fail because ivn does not contain any values
### (compare to ivn.vals above).
## Not run: 
data.ivn <- ccdrAlgorithm::generate_mvn_data(graph = gr,
                                             n = 100,
                                             params = gr.params,
                                             ivn = ivn,
                                             ivn.rand = FALSE)

## End(Not run)

ccdrAlgorithm documentation built on April 12, 2022, 9:06 a.m.