Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Reproducible Path Diagrams (Graphviz and DOT in Stata)

    I have written a package named `diagram` (installable through GitHub only, at the moment). Earlier I wrote the same package for R, called DOT.

    `diagram` can:
    1. Render `Graphviz` diagrams in Stata
    2. Render path diagrams written in `DOT` markup language
    3. Change the resolution of the diagram to any size
    4. Export the graph as raster or vector graphics as pdf, png, bmp, gif, and jpeg
    The package can have many applications in drawing reproducible SEM diagrams, visualizing networks data, etc.

    The package includes 3 example programs for generating a dynamic diagram for the `sem` and `gsem` commands in Stata, generating a dynamic diagram from the dataset, and also visualizing ado programs based on called programs (i.e. nested ado-programs).

    If you are interested, send me an email and I will also share the manuscript I have written for Stata Journal (haven't submitted it yet, want to read it again in a week or two, so I appreciate feedback and suggestions).

    Here is an example of rendering a DOT file that exists in the example directory:

    Code:
    diagram using "./examples/PATH1.txt", export(./examples/example1.png)
    Click image for larger version

Name:	example1.png
Views:	1
Size:	504.2 KB
ID:	1351978
    ——————————————
    E. F. Haghish, IMBI, University of Freiburg
    [email protected]
    http://www.haghish.com/

  • #2
    E.F.: Sounds very interesting and useful.

    Footnote: Markup

    Code:
     
    `  `
    has no special effect on Statalist, unlike (e.g.) Stack Exchange.

    Comment


    • #3
      Nick I guess my Markdown addiction is public now

      PS. What I am excited about is the semdiagram.ado (included in the package) which reproducible generates reproducible "simple" SEM model. This is really the way to go, but the layout should be further developed and fine-tuned for statistical models.

      ——————————————
      E. F. Haghish, IMBI, University of Freiburg
      [email protected]
      http://www.haghish.com/

      Comment

      Working...
      X