You can get the lastest versions from Github:
net install pystacked, /// from(https://raw.githubusercontent.com/aahrens1/pystacked/main) replace
Please check for updates on a regular basis.
pystackedrequires at least Stata 16 (or higher), a Python installation and scikit-learn (0.24 or higher). Python and scikit-learn are available for free.
You can also install from SSC, but note that we update the SSC version less regularly:
ssc install pystacked
Install old versions #
To install an old version:
net install pystacked, from(https://raw.githubusercontent.com/aahrens1/pystacked/vXXX) replace
XXX is replaced with one of the archived branches.
Setting up Python #
If you haven’t set up Python for Stata, type
help python and check this Stata blog post for how to set up Python for Stata on your system.
In short, you can either install Python manually (e.g. from www.python.org/) or use a distribution such as Anaconda. Anaconda is in most cases the easier method.
After you have installed Python, you might also need to tell Stata where your Python installation is located. You
can do this using
python set exec. Note that you will usually have more than one Python installation on your system, since Python is shipped with all common operating systems (yet, usually an old version).
python search will show all
Python installations Stata can find.
Updating scitkit-learn #
pystacked requires scikit-learn 0.24 or higher. You can check your scikit-learn version using:
. python: import sklearn . python: sklearn.__version__
If you use Anaconda, you can use the Anaconda Navigator (or the
conda command line tool) to update packages. Otherwise you can use
pip (see here).
For example, if your Python installation is located in /usr/local/bin/python3.9, you could update scikit-learn by typing
/usr/local/bin/python3.9 -m pip install -U scikit-learn
into the terminal, or directly in Stata (restart required):
. shell /usr/local/bin/python3.9 -m pip install -U scikit-learn
Offline installation #
If you want to use
pystacked in an offline environment,
we recommend to download the packages from the Github repository. Click the green button “Code”
and “Download ZIP”. Then run
net install as above but
should refer to the downloaded and unzipped repository folder.
Verify installation #
To check that the packages were installed correctly, type e.g.
which requires the user-written package whichpkg.
We recommend to add this to your log files to facilitate reproducibility.