Network_comparison

This repository serves as a research archive for the mini-project "Comparison of Gaussian graphical models (GGM) and Directed Cyclic Graph (DCG) Models as Causal Discovery Tools", which is an extension of my Master's thesis (mid-report PDF here). This repository contains all necessary files to reproduce the original simulation study and the follow-up analysis.

Contents

The file structure is shown as below.

├── LICENSE
├── Network_comparison.Rproj
├── code
│   ├── R
│   │   ├── CCD_fnc.R
│   │   ├── data_generating_fnc.R
│   │   ├── dsep_fnc.R
│   │   ├── equivset_fnc.R
│   │   ├── eval_metric_fnc.R
│   │   ├── plot_fnc.R
│   │   ├── searchAM_KP_fnc.R
│   │   └── variation_fnc.R
│   ├── empircal_example.R
│   ├── simulation.R
│   └── variation_DCG.R
├── data
│   ├── McNally.csv
│   ├── equiv4p.RData
│   ├── equiv4p_high.RData
│   ├── equiv5p.RData
│   ├── equiv5p_high.RData
│   ├── equiv6p.RData
│   └── equiv6p_high.RData
└── manuscript
    ├── apa.csl
    ├── img
    │   ├── CCDoverallsummary.png
    │   ├── CCDsummary.png
    │   ├── degreevariation.png
    │   ├── densityvariation.png
    │   └── truemodels.png
    ├── follow-up.html
    ├── follow-up.qmd
    ├── original_simulation.html
    ├── original_simulation.qmd
    ├── references.bib
    ├── style.css
    └── style2.css

5 directories, 35 files

CODE

• code folder contains all the code used for the original simulation study and follow-up analysis.
• Three R scripts contains the code for simulating models in the original study and creating figures presented in the follow-up report.
  • simulation.R: code to simulate all six different models.
  • variation_DCG.R: code to create the density plots (Figure 2) and degree centrality plots (Figure 3) in the follow-up report.
  • empircal_example.R: code to create the PAG (Figure 4) and GGM (Figure 5) on empirical data in the follow-up report.
• R folder contains all supporting functions required to run the R scripts mentioned above.

DATA

• data folder contains six equivalence classes of directed cyclic graphs (DCG) from each of the simulated models saved as .Rdata.
• Also, an example empirical data McNally.csv can be found in this folder.

MANUSCRIPT

• original_simulation.html is the manuscript for the main analysis.
• original_simulation.qmd renders the main manuscript.
• follow-up.html is the manuscript for the follow-up analysis.
• follow-up.qmd renders the follow-up manuscript.
• other supporting files/folders:
  • img: img folder contains all necessary image files to render the manuscripts.
  • references.bib: BibTex for references cited in the manuscripts.
  • apa.csl: csl file to format the references as per APA guidelines.
  • .css files: for the minor styling and layout of the manuscripts.

RECOMMENDATION

1. Check the manuscripts. (original_simulation.html and follow-up.html)
2. To reproduce the original simulation study, either check the code presented in original_simulation.html or check the quarto document that contains all the code.
3. To reproduce the follow-up analysis, run the following R scripts: variation_DCG.R and empircal_example.R for the first sub-analysis and the second sub-analysis, respectively.
4. For any further information, see the detailed description attached to each R script.

---

For any help with the files in this archive, please contact Kyuri Park (k.park@uu.nl).