miniconda

Here's a crash-course in making your own python toolkit via miniconda:

Although it is recommended to install the full anaconda system via https://www.continuum.io/ this is an alternative if you want to cherry pick your modules and start with a very small python system. It's easy super easy to install a new python this way, since you can do it with just a few shell commands.

For miniconda, head over to http://conda.pydata.org/miniconda.html and grab the latest installer and follow instructions for your architecture.

Here is an example for linux 64 bit:

   wget https://repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh

   bash Miniconda3-latest-Linux-x86_64.sh

    (answer some obvious questions)
    I installed it in:   /data2/teuben/python/miniconda3
    This modified my PATH as follows in the .bashrc file:
    export PATH="/data2/teuben/python/miniconda3/bin:$PATH"

   python --version                         (3.5.2)
   conda install ipython
   ipython --version                        (4.2.0)
   conda install numpy scipy matplotlib
   conda install yt
   conda install jupyter
   conda install astropy pandas scikit-learn glueviz
   conda install anaconda-client

Now you can issue the command

   jupyter notebook

and in your browser you will now see python notebooks (ipynb files) if you are in the right directory. Since this is a lot of typing, an alias

   alias jn="jupyter notebook"

would seem appropiate here.