[PYTHON] Build a PyData environment for a machine learning study session (January 2017)

Prerequisites

We have decided to hold a study session on the subject of Python Machine Learning Programming (Impress). Since Python 3.6 has also been released, I will summarize how to build the environment again.

The environment construction performed here is as follows.

• VirtualBox + Vagrant
• Ubuntu 16.04 LTS
• Python 3.6 --PyData related (NumPy / Pandas / scikit-learn (sklearn))

The following is assumed to have been installed.

• VirtualBox (5.0.30)
• Vagrant (1.9.1)

$ VirtualBox --help
Oracle VM VirtualBox Manager 5.0.30

$ vagrant version
Installed Version: 1.9.1
Latest Version: 1.9.1

VirtualBox + Vagrant

Prepare a folder for Vagrant. This time, I will study machine learning, so prepare a folder called ~ / vagrant / ml-study and prepare the environment.

$ mkdir ~/vagrant/ml-study

Addition of Vagrant Box

$ vagrant box add ubuntu/xenial64 https://atlas.hashicorp.com/ubuntu/boxes/xenial64/versions/20170104.0.0/providers/virtualbox.box

Start Vagrant

$ vagrant init ubuntu/xenial64
$ vagrant up

Vagrant settings (edit Vagrantfile) Enable config.vm.provision at the bottom and write:

...
  config.vm.provision "shell", inline: <<-SHELL
    apt-get update
    sudo apt-get install -y build-essential libssl-dev libxml2-dev libxslt1-dev libbz2-dev zlib1g-dev python-setuptools python-dev libjpeg62-dev libreadline-gplv2-dev
    sudo apt-get install -y libblas-dev liblas-dev liblapack-dev gfortran libfreetype6-dev
  SHELL
...

Reflection of settings

$ vagrant provision

OS related packages

Since the necessary packages are described in Vagrantfile, it will be installed on the vagrant side.

Python 3.6

$ mkdir ~/tmp
$ sudo mkdir /opt/python36
$ sudo chown ubuntu /opt/python36

$ cd tmp
$ wget https://www.python.org/ftp/python/3.6.0/Python-3.6.0.tgz
$ tar zxvf Python-3.6.0.tgz
$ cd Python-3.6.0/
$ ./configure --prefix=/opt/python36
$ make && make install

$ cd ~
$ /opt/python36/bin/python3 -m venv python
$ source python/bin/activate

Python related modules

$ pip install numpy scipy pandas scikit-learn matplotlib
$ pip install ipython[notebook]
$ pip install seaborn
$ pip install pyprind

When using after setting

$ cd ~/vagrant/ml-study
$ vagrant up
$ vagrant ssh

$ cd ~
$ source python/bin/activate
$ jupyter notebook

When stopped

$ vagrant halt

jupyter notebook settings

(Added on January 8, 2017)

When I try to access Jupyter notebook directly from the host machine with a browser, it is rejected due to IP restrictions. IP restrictions can be lifted by making the following settings.

$ source python/bin/activate
$ jupyter notebook --generate-config

~/.jupyter/jupyter_notebook_config.py

c.NotebookApp.ip = '*'