Click here for Documentation.
These are installation instructions for using py-tetrad in RStudio to run algorithms in Tetrad from R.
We have worked out and tested these instructions using RStudio on a Mac; they have been tested (and adjusted) for Windows by end-users. We have not tested them on Linux yet.
Some steps of this installation can be done automatically in the future using a Java package or a Docker container. We apologize for the current complexity.
For very verbose instructions, see here.
If you're using a version lower than 3.7, update it.
If you don't know which paths you want to use for these already, open a Terminal window and type:
which java
which python
Remember these paths for the steps below.
(4) On a Mac (or maybe Linux too?) Set JAVA_HOME to your Java installation path in the file "~/.Renviron".
On Windows, once you install Java in step (1), this step is unnecessary! Lucky you!
On Mac, in a text editor, check to see if you have a file called .Renviron in your home directory, ~; if not, create one. In this file, type this line, for example--use the path to the Java JDK on your machine that you found in step (3):
JAVA_HOME = /Users/[username]/Library/Java/JavaVirtualMachines/[JDK name]/Contents/Home
Save this file as .Renviron in your home directory, ~. Again, this should be the path to the .../Home directory of your JDK.
Then, when you open RStudio below by double-clicking on its icon after following step 4, step 5 should work. If not, come back to this step and double-check your work.
pip install causal-learn
pip install JPype1
git clone https://github.com/cmu-phil/py-tetrad/
To test this installation, type
cd py-tetrad/pytetrad
python run_continuous.py
This last command should cause various algorithms to run in Tetrad and print out result graphs.
(Here are docs for JPype and git.)
Open RStudio, and inside RStudio, type the following:
install.packages("reticulate")
install.packages("DiagrammeR")
install.packages("psych")
reticulate::use_python("[...path...to...your...Python...installation...]")
For the last command, this may work:
reticulate::use_python(Sys.which("python"))
This Python path was found in step (3). "DiagrammR" is used in the sample R scripts to render graphs in the RStudio viewer window; if you're running the scripts in command-line R outside of RStudio, it will not help. On Windows, "DiagrammR" must be installed with devtools for the latest R versions. The "psych" package is used in the examples to display scatterplots and histograms of the data.
This only needs to be done once.
(Here are the docs for RStudio, the Reticulate package in R, the DiagrammeR package in R, and the psych package in R.)
(7) Finally, open up RStudio, and in RStudio, open one of the example R scripts in the py-tetrad repository and run it.
In RStudio, open the file 'py-tetrad/pytetrad/R/sample_r_code2.R', for example.
Once you've loaded it, adjust the path to the 'py-tetrad/pytetrad' directory if it isn't right already--this line:
setwd("~/py-tetrad/pytetrad")
source("R/sample_r_code2.R")
Select all lines in the file by typing control-A.
Then click the Run button. That should run FGES on the example file, display the result in the Viewer window using Graphviz, and display a plot matrix of the scatterplots and histograms of the variables in the Plot window.
We assume you know many other things you can do with data in R if you're an R user.