subgroupsem is an R package for discovering interesting subgroups with distinct sets of parameters in structural equation models (SEM). It can be installed via GitHub.
subgroupsem is currently not on CRAN. The development version of subgroupsem can be installed directly from this GitHub repository using the additional package devtools. Under Windows, please make sure Rtools (http://cran.r-project.org/bin/windows/Rtools) is installed and no older version of subgroupsem is currently loaded. Also, make sure that the packages lavaan (structural equation modeling) and reticulate (python interface) are installed.
# install reticulate
install.packages("reticulate")
# install lavaan
install.packages("lavaan")
# install devtools
install.packages("devtools")
# install subgroupsem
devtools::install_github("langenberg/subgroupsem")
Please make sure that you installed a python distribution. Python is not included in the subgroupsem package. You can either install python separately or use miniconda which can be installed using the reticulate package:
reticulate::install_miniconda()
For more information, we would like to refer the reader to reticulate website.
Lastly, you need to install the python module pysubgroup. You can either do this yourself using reticulate:
reticulate::py_install("pysubgroup==0.7.2", pip = T)
Or you can have subgroupsem install the python module for you. When you run the subgroupsem() function (or one of the wrapper function, e.g., subsem()) and pysubgroup is not installed, you will be asked to install the module:
Python module 'pysubgroup' not installed. Do you want to install now? (y/n)
Simply confirm by typing y.
The main function of the package is subgroupsem(). The documentation of this function helps to introduce you to its functionality.
If you decided to use miniconda, do not forget to run the command use_miniconda() after you loaded subgroupsem. This has to be done again every time you loaded subgroupsem.