Build an environment for machine learning using Python on MacOSX

This is a memo for building a development environment for reading the following books.

Book: Scraping & Machine Learning Development Techniques with Python http://www.socym.co.jp/book/1079

Homebrew installation

First of all, install the package manager "Homebrew" for Mac OS X. http://brew.sh/index_ja.html

$ /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

On the way, enter the sudo password.

pyenv

Next, install the Python version manager "pyenv".

You can change the environment (installed packages) for each Python version.

$ brew install pyenv

Set environment variables.

$ echo 'export PYENV_ROOT="$HOME/.pyenv"' >> ~/.profile
$ echo 'export PATH="$PYENV_ROOT/bin:$PATH"' >> ~/.profile
$ echo 'eval "$(pyenv init -)"' >> ~/.profile
$ . ~/.profile

Check the versions and packages that can be installed using pyenv.

$ pyenv install --list

Install the latest Python 3.6.0 and anaconda 3-4.2.0.

$ pyenv install 3.6.0
$ pyenv install anaconda3-4.2.0

Check the installed version.

$ pyenv versions

Switch the version as follows. (Change global to local if you want to apply only to a specific directory)

（python3.6.Switch to 0)
$ pyenv global 3.6.0
$ pyenv rehash
$ python --version
Python 3.6.0
$ pyenv versions
  system
* 3.6.0 (set by /Users/hisashi/.pyenv/version)
  anaconda3-4.2.0

（anaconda3-4.2.Switch to 0)
$ pyenv global anaconda3-4.2.0
$ pyenv rehash
$ python --version
Python 3.5.2 :: Anaconda 4.2.0 (x86_64)
$ pyenv versions
  system
  3.6.0
* anaconda3-4.2.0 (set by /Users/hisashi/.pyenv/version)

Package management

Python package management is done with "pip". In addition, Anaconda package management is performed by "conda" and "pip".

editor

I think Atom is a good editor. https://atom.io

PhantomJS and Selenium

It is a setting of scraping via a browser (Chapter 2, Section 2). We will use "anaconda 3-4.2.0" which is often used in machine learning.

$ pyenv global anaconda3-4.2.0
$ pyenv rehash

The book introduces the environment construction with Ubuntu + docker, but with MacOSX you can build the environment as follows.

$ brew install phantomjs
$ pip install selenium
$ pip install beautifulsoup4

Install TensorFlow

I will continue to use "anaconda 3-4.2.0".

In the book, the following command is described, but an error occurs and it stops halfway.

$ pip install --upgrade https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-0.12.1-py3-none-any.whl

You should add the "-I setuptools" option to avoid errors. http://datalove.hatenadiary.jp/entry/python/anaconda/install-tensorflow-into-anaconda-environment

$ pip install --upgrade -I setuptools https://storage.googleapis.com/tensorflow/mac/cpu/tensorflow-0.12.1-py3-none-any.whl