equSA (version 1.1.5)

p_learning: Construct Bayesian Network based on p-learning algorithm.

Description

Construct Bayeisan network for general types of random varaibles based on p-learning algorithm.

Usage

p_learning(data, gaussian.index, binary.index, poisson.index, 
alpha1 = 0.1, alpha2 = 0.02, alpha3 = 0.02)

Arguments

data

The data matrix, of dimensions nxp. Each row is an observation vector.

gaussian.index

The index vector of continuous nodes. The default value is NULL.

binary.index

The index vector of binary nodes. The default value is NULL.

poisson.index

The index vector of poisson nodes. The default value is NULL.

alpha1

The significant level of step(a) of p-screening method. The default value is 0.1.

alpha2

The significant level of step(c) of p-screening method. The dafault value is 0.02.

alpha3

The significant level of solving Markov Blankets. The default value is 0.02.

Value

A list of one object.

PDAG

The derived partial directed acyclic graph.

Details

This is the function that implements the p-learning algorithm.

References

Suwa, Xu and Faming, Liang (2017). Learning High-Dimensional Bayesian Networks for General Types of Random Variables. Submitted to Biometrika.

Examples

# NOT RUN {
 #library(equSA)
 #data(mixed3000)
 #pdag3000 <- p_learning(data =mixed3000$data, gaussian.index = 
 #mixed3000$gaussian.index,binary.index <- mixed3000$binary.index)$PDAG



# }