Instructions to build Python environment

Linux, using Python 2.7, system packages, tested on Ubuntu 16.04 Vagrant box

sudo apt-get update
sudo apt-get install python python-nose-parameterized python-numpy python-sklearn python-pip python-bs4 python-pandas
sudo pip install --upgrade pip
sudo pip install theanets

or use provided Vagrantfile to setup VM.

Linux, using Python 2.7, virtualenv

Install system packages

sudo apt-get install python python-pip python-virtualenv python-tk

Install remaining packages in virtualenv

virtualenv venv27
venv27/bin/pip install -r requirements27.txt

Linux, using Python 3.5, system packages, tested on Ubuntu 16.04 Vagrant box

sudo apt-get update
sudo apt-get install python3 python3-nose-parameterized python3-numpy python3-sklearn python3-pip python3-bs4 python3-pandas
sudo pip3 install --upgrade pip
sudo pip3 install theanets

or use provided Vagrantfile to setup VM.

Linux, using Python 3.5, virtualenv

Install system packages

sudo apt-get install python3 python3-pip python3-virtualenv python3-tk

Install remaining packages in virtualenv

virtualenv -p `which python3` venv35
venv35/bin/pip3 install -r requirements35.txt

MS Windows, using Python 2.7, anaconda

Download from continuum.io and install Anaconda for Python 2.7 (tested for Anaconda 4.4 on Windows 10)

The Anaconda Python installation contains required packages for all chapters except Artificial neuron networks.

For the last one, we need to install Theano and nose-parameterized by Conda and then theanets by pip.

In Anaconda prompt:

conda install nose-parameterized theano
pip install theanets

MS Windows, using Python 3.6, anaconda

Download from continuum.io and install Anaconda for Python 3.6 (tested for Anaconda 4.4 on Windows 10)

The Anaconda Python installation contains required packages for all chapters except Artificial neuron networks.

First try the procedure for Python 2.7, if it does not work (due to version incompatibility between pygpu and theano, perhaps) then the following.

Install theano with dependencies and nose-parameterized by conda, deinstall pygpu and theano from conda, install theano and theanets by pip.

In Anaconda prompt:

conda install nose-parameterized theano
conda uninstall pygpu
pip install theano
pip install theanets

Run tests in command line

Run from the directory of a chapter (not repository root directory).

python -m unittest discover tests

or

../venv35/bin/python3 -m unittest discover tests

or

../venv27/bin/python -m unittest discover tests

Run tests in PyCharm

If you PyCharm project is the repository, then mark directory of the chapter as sources root (in Project panel, in the context menu of directory "Mark Directory As" -> "Sources Root").

For the single test in Project panel, in the context menu of file "Create Unittests in test_corpus_parser" and make sure that working directory is "something/hidden_markov_model", but not "something/hidden_markov_model/tests".

For all tests do the same in the context menu of the "tests" directory.