Installing Jupyter on a Mac using Homebrew

Here's how to install Jupyter with python and R integration on OSX using Homebrew.


Step 0 - Install Homebrew

Read this tutorial on how to install Homebrew on OS X.


Step 1 - Install pyenv

Mac Os X come with Pythong 2.7 pre-installed but many Machine Learning packages are progressing to Python 3.x. Therefore, it's recommended you start using Python 3 and the best way to do that is to first install pyenv version manager. This will allow you to install any version of Python you'd like.

First update Homebrew package manager.

brew update && brew doctor

Install pyenv version manager.

brew install pyenv

Step 2 - Install Python

Install Python 3.x using pyenv. You can see a list of version from the Python website.

pyenv install -l | grep -ow [0-9].[0-9].[0-9]
pyenv install 3.x.x

Double check your work.

pyenv versions

You'll also need to configure your ~/.bash_profile.

echo 'eval "$(pyenv init -)"' >> ~/.bash_profile 

If you need more help with pyenv, I suggest you reading my other article titled Installing multiple versions of python on OS X using Homebrew


Step 3 - Set Python to Local or Global

If you only want to use Python 3.x for a specific project, then cd to your specific directory and type:

pyenv local 3.x.x

If you'd prefer to just have Python 3.x installed globally throughout your operating system, then type:

pyenv global 3.x.x

Step 4 - Install Jupyter

Jupyter is an acronym for Julia, Python and R but these days, other languages are also included such as Ruby.

brew install jupyter

Step 5 - Start Jupyter

Now it's time to start the jupyter notebook.

jupyter notebook

Step 6 - Done!

You're Done Installing Jupyter. Please proceed if you now want to use R language with Jupyter.


Troubleshooting

~/.bash_profile

If you're struggling to get configure bash profile, this has also worked for a few different developers.

export PYENV_ROOT=/usr/local/opt/pyenv
eval "$(pyenv init -)"

Resources

Although it takes a little bit of extra work to get R working with Jupyter, it's totally worth it. Otherwise, if you want to import data, clean it, structure it and process it, you'll have to have to learn all of these other Python libraries to do what R does natively.

  • NumPy supports scientific computing and linear algebra support.
  • Pandas provide data frames which make it easy to access and analyze data. This is a data manipulation tool.
  • Mat plotlib is a 2D publication library that produces high quality graphics.
  • Scikit-learn's purpose is to support machine learning and therefore it's used for many of the tasks performed routinely in machine learning. A few key features are:
    • It works well with the libraries stated above.
    • It helps integrate the algorithms we will use for predictive models.
    • Methods that will help us pre-process data.
    • Methods for helping us measure the performance of our models.
    • Methods for splitting data into test sets
    • Methods for pre-processing data before training.
    • Methods for creating trained models, tuning models and identifying which features within the models are important.