Prepare a programming language environment for data analysis

I talked about Preparing a computer environment for data analysis before, but even if you only have a computer, you can't talk about it without a programming language environment. ..

Today I'll show you how to install Python and its related libraries.

Python build

The latest version 3.4.1 was released on 5/18.

Reasons to build your own

Python setup methods include virtual environments for languages such as virtualenv, apt and [brew. Many people use a package management system like [http://brew.sh/).

I recommend building your own programming languages that you often use. The main reasons are as follows.

1. Can be built using the latest source code that has not been released yet
2. If there is a bug in the language or the behavior you want to change, you can fix it and build it yourself.
3. Make it easier to participate in language development
4. You can follow common steps on many platforms without being locked into a specific package management system or virtual environment middleware.

Build installation destination

Also, by standardizing the build and installation destination, it will be easier to understand when uninstalling or switching versions.

The author specifies the directory for most products as follows.

/opt/[Product name]/[version]

It also creates a symbolic link under / opt / [product name] / called current and links it to the version you want to use.

For example:

$ ls -la /opt/python/
drwxr-xr-x 3.3
drwxr-xr-x 3.4
drwxr-xr-x trunk
lrwxrwxrwx current -> 3.4

$ ls -la /opt/ruby/
drwxr-xr-x 1.9.3
drwxr-xr-x 2.0
drwxr-xr-x 2.1
lrwxrwxrwx current -> 2.1

This makes it easy to switch between versions coexisting with older versions, and even uninstalling can be easily removed with rm -rf.

Installation script

Also, builds use shell scripts rather than manual ones.

For example, to install Python, use this script. https://github.com/ynakayama/tagokura-python/blob/master/installer/install_python.sh

At startup, specify the version in the first argument and the installation destination in the second argument, as described in the comments.

~/install_python.sh 3.4.1 /opt/python/3.4

If you can build by just launching a shell script, you don't have to repeat the manual work when installing on a different host or when building a newer version.

Recently, automation frameworks such as chef have become popular, but shell scripts have long been traditional, so that's right. You don't have to worry about it becoming obsolete easily. It can be used even in the smallest environment that has been around for a long time, and it is easy to check the behavior by looking at the contents when something goes wrong.

Install the pip package

It is recommended to install the pip package as well as the main unit with a shell script.

It is a good idea to put all the necessary packages together by referring to this script. https://github.com/ynakayama/tagokura-python/blob/master/installer/install_pip.sh

If you use AWS, you should also include the AWS Command Line Interface.

Summary

It is convenient to standardize and automate the installation work of programming languages used in the computer environment. When using in a distributed environment, specify the version and install using the same procedure so that there is no difference between the versions.