Mac Machine Learning Research & Dev Setup

Update #1: Post updated to reflect macOS’ move to zsh and the move to Python3 (09/06/2020).

Every year I do a clean install of the latest macOS, to mostly get rid of the things I accumulated throughout the year that I no longer use and force myself to do some housekeeping. Whether after a clean install or just getting my hands on a new Mac, I find myself going through the same process every time. In this post, I document this workflow for my future self and hopefully for the benefit of others.

Before

Back in the day, backing up all of my data just before a clean install was a painful process but thanks to iCloud, which now takes care of backing up my bookmarks, passwords, etc., it is now a straightforward process. All I have to do is copy my files in ~/$USER to an external hard drive. P.S. I do this in addition to the Time Machine backup just in case!

Format

Prepare a bootable macOS installer, erase the main drive, and install the OS. A nice guide for this can be found here.

Settings & Software

First, copy your files back and change the trackpad, keyboard, user, etc. settings back to your liking. Don’t forget to turn off font smoothing. I also have a bunch of dot files that contain paths, aliases, etc. to make things easier. I keep these files in this Github repository, some of these files might be useful.

Also, install your favorite editor, TeX/iWork/Office, and other software you regularly rely on, as well as, themes and colors.

The remainder of the workflow will be run from the terminal.

Command Line Tools

Second, we’ll install the command line tools:

xcode-select --install 

Don’t forget to accept the license agreement after installation before moving on if you’re installing Xcode.

Package Manager

Install a package manager, Homebrew (or Macports):

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

Some Standard Packages

A version control system - Git (or preferred alternative):

brew install git
git config --global user.name "Your Name Here"
git config --global user.email "your_email@youremail.com"
git config --global credential.helper osxkeychain

A minimal .gitconfig for Git settings.

A C compiler - GCC; this is also a dependency for later packages:

brew install gcc

Python3:

brew install python

The world’s best text editor :) - Vim:

brew install vim

A minimal .vimrc for some sensible vim settings.

It’s useful to have Ruby and Gem installed:

brew install ruby

This website is built with jekyll:

gem install --user-install bundler jekyll

Scientific Stack

I used to follow a somewhat involved process. Now, installing the usual suspects NumPy, SciPy, MatPlotLib, Jupyter, and Virtualenv is as simple as:

python3 -m pip install numpy
python3 -m pip install scipy
python3 -m pip install matplotlib
python3 -m pip install jupyter
python3 -m pip install virtualenv

Other packages should probably be installed within a virtual environment.

(While you’re at it, do consider donating :) to the open-source community.)

Finally, the machine learning toys - scikit-learn and PyTorch:

python3 -m pip install scikit-learn
python3 -m pip install torch

End

That’s all folks! I find this to be a minimal setup to get up and running. Other tools can be installed later when needed and probably within the project scope.

Citation:

@misc{fayek2018,
  title   = "Mac Machine Learning Research & Dev Setup",
  author  = "Haytham M. Fayek",
  year    = "2018",
  url     = "https://haythamfayek.com/2018/01/14/mac-ml-setup.html"
}