# Documentation DocumentationWe use a combination of sphinx and Jupyter notebooks for the documentation.Jupyter notebooks should be used for longer, self-contained examples demonstratinga topic.Sphinx is nice because we get the tables of contents and API documentation.## Build ProcessBuilding the docs requires a few additional dependencies. You can get mostof these with```bash pip install-e .[docs]```From the root of the project.Some of the examples rely on `rpy2` to execute R code from the notebooks.It's not included in the setup requires since it's known to be difficult toinstall.To generate the HTML docs, run ``make html`` from the ``docs`` directory.This executes a few distinct builds1. datasets2. notebooks3. sphinx# Notebook BuildsWe're using `nbconvert` to execute the notebooks, and then convert themto HTML. The conversion is handled by `statsmodels/tools/nbgenerate.py`.The default python kernel (embedded in the notebook) is `python3`.You need at least `nbconvert==4.2.0` to specify a non-default kernel,which can be passed in the Makefile.