Python for Data Analysis: Setup

Reading over the documentation for Enthought Deployment Manager, it looks like the better way to go about this, instead of downloading a big set of dependencies, will be to load dependencies into my working environment as they become necessary.

Setting Up Enthought Deployment Manager (EDM)

EDM has authentication options for users with commercial accounts, including simple auth (not recommended) and API tokens (recommended). I don’t have such an account, so I’ll worry about that in the future. If that’s you, I can’t help you, but check out edm help configuration and edm configure.

Get Victoria’s stories in your inbox

Join Medium for free to get updates from this writer.

Remember me for faster sign in

Instead, I’ll get straight to setting up my environment. EDM handles a really ambitious amount of stuff, from dependency resolution to environment management. I’m a little skeptical of this approach, but we’ll go with it for now.

edm environments create python_for_data_analysis

That ran uneventfully. It assumed Python 3.6 in the absence of me providing a version, which is fine with me. Next we activate it:

edm shell --environment python_for_data_analysis 

There was a small wrinkle here, as I’m using fish, an unsupported shell. EDM wasn’t happy about that, but I’ll take my chances. I did a quick check after activating the environment:

which python3
> /home/victoria/.edm/envs/python_for_data_analysis/bin/python3

Looks good! Now, I’ll load up this environment with Kinney’s recommendations and we can get on to the fun stuff. Here’s the packages I added with edm install initially:

• NumPy
• pandas
• SciPy
• IPython
• matplotlib

ERROR! “No module named ‘PySide’”

Starting up IPython for the first time with ipython --pylab , I got a neat error about my lack of the PySide module. I don’t know what that is. A quick Google search reveals it’s an alternative to Python’s standard tkinter library for developing GUIs with Python.

Looks like we can’t get PySide2 with edm install , but which pip points to the right environment, so I’ll give it a go with pip .

Nice! That seems to have done it. IPython starts uneventfully when I run ipython --pylab . Looks like we should be good to move on, at last, to the fun stuff!