# SimGraDat: Simulate Incomplete Data for Gaussian Graphical Models In equSA: Learning High-Dimensional Graphical Models

## Description

Simulate compeletely missing at random (CMAR) data with a band structure, which can be used in GraphIRO(data,...) for estimating the structure of the Gaussian graphical network.

## Usage

 1 SimGraDat(n = 200, p = 100, type = "band", rate = 0.1) 

## Arguments

  n  Number of observations, default of 200.  p  Number of covariates, default of 100.  type  type=="band" which denotes the band structure, with precision matrix C_{i,j}=≤ft\{\begin{array}{ll} 0.5,&\textrm{if $≤ft| j-i \right|=1, i=2,...,(p-1),$}\\ 0.25,&\textrm{if $≤ft| j-i \right|=2, i=3,...,(p-2),$}\\ 1,&\textrm{if $i=j, i=1,...,p,$}\\ 0,&\textrm{otherwise.} \end{array}\right.  rate  Missing rate, the default value is 0.1.

## Value

  data  nxp Gaussian distributed data with missing.  A  pxp adjacency matrix used for generating data.

## Author(s)

Bochao Jiajbc409@gmail.com and Faming Liang

## References

Liang, F., Jia, B., Xue, J., Li, Q., and Luo, Y. (2018). An Imputation Regularized Optimization Algorithm for High-Dimensional Missing Data Problems and Beyond. Submitted to Journal of the Royal Statistical Society Series B.

## Examples

 1 2 library(equSA) SimGraDat(n = 200, p = 100, type = "band", rate = 0.1) 

equSA documentation built on May 6, 2019, 1:06 a.m.