2. Setup

Pmdarima depends on several prominent python packages:

2.1. Install from PyPi

Pmdarima is on pypi under the package name pmdarima and can be downloaded via pip:

$ pip install pmdarima

Pmdarima uses Cython, which means there is some C source that was built in the distribution process. To ensure the package was built correctly, import the following module in python:

from pmdarima.arima import auto_arima

If you encounter an ImportError, try updating numpy and re-installing. Outdated numpy versions have been observed to break the pmdarima build.

2.2. Install from Conda

Pmdarima is on conda under the channel conda-forge and package name pmdarima. It can be downloaded via conda like so:

$ conda config --add channels conda-forge
$ conda config --set channel_priority strict
$ conda install pmdarima

Note: We do not maintain our own Conda binaries, they are maintained at https://github.com/conda-forge/pmdarima-feedstock. See that repo for further documentation on working with Pmdarima on Conda.

2.3. Build from source

If you’d like to install a development or bleeding edge version of pmdarima, you can always build it from the git source. First clone it from Git:

$ git clone https://github.com/alkaline-ml/pmdarima.git
$ cd pmdarima

Building the package will require gcc (unix) or a Windows equivalent, like MinGW. To build in development mode (for running unit tests):

$ python setup.py develop

You can also use the Makefile if you’re on a posix machine:

$ make develop

Alternatively, to install the package in your site-packages:

$ python setup.py install

Or, with the Makefile:

$ make install