"Collaboratool" is a project for building, integrating, and deploying tools that support portable, reproducible data science. It is currently a collaboration between folks at UC Berkeley in the D-Lab, EECS, Statistics, and the iSchool.

We got started with thinking about how to deploy virtualized containers that provide things like IPython notebooks through the web. We were very inspired by jiffylab. We will likely include some variant of this project in the collaboratool suite (or "shed" as we like to call it).

Currently, the bsd2 directory contains work towards a "Berkeley Standard Data Science Distribution."

Shortly, we'll be putting up a website using github pages, and some of this info will migrate there. Stay tuned. For now, here are some lists:

Virtualized IPython notebooks:

• jiffylab was the first project we saw that did this. It relies on Docker as a primary enabling technology.
• ipydra is the current home of work begun in ipython-hydra. It doesn't appear to use the same level of virtualization
• ipython-dokku, perhaps unsurprisingly, uses dokku to achieve similar things. SimpliPy uses this to deploy to Digital Ocean.
• notebookcloud does something similar, but is restricted to EC2.
• Continuum Wakari and picloud offer proprietary solutions, though both provide for a certain amount of customization. Last time Dav checked, Waraki customization was more limited and less reliable than picloud (though it requires less admin knowledge).

Similarly, [RStudio Server](RStudio Server) runs through a web interface as well, but we have yet to identify a solution like jiffylab or the others above.