News for hackers. Stuff that matters.

A comment on Data Science Toolbox.

Re:Data Science Toolbox by jmount on Wednesday June 04, 2014 @02:45AM (#7844311)
I don't get it. DST installs 3 tools: R, Python, and DST itself. R and its packages are VERY easy to install, Python and its packages are pretty easy to install, and you don't need DST until you choose to install DST. It's not like they are putting a large hard to configure stack (like Hadoop/Mahout/Pig/Hive/Spark ...) for you.
[ Parent | on HN ]
Re:Data Science Toolbox by jeroenjanssens on Wednesday June 04, 2014 @06:17AM (#7844824)
Author here. Although the software contained in the base is indeed relatively easy to install, you'd be surprised how often people struggle installing Python and its packages on Mac OS X or Microsoft Windows!

The initial reason I created DST was that I'm currently writing a book called Data Science at the Command Line. Installing the command-line tools discussed in this book is unfortunately not straightforward. I wanted to offer potential readers a Vagrant box (with a shell provisioner) that installs these command-line tools. Similar to what Matthew Russell has done for his book Mining the Social Web [1]. This blog post [2] provides some more explanation and compares four different solutions.

The software packages that you mention sure are exciting! If there is an author or teacher that wants to teach these to her readers or students, then we should create a bundle (i.e., Ansible playbook) for them!

[1] https://github.com/ptwobrussell/Mining-the-Social-Web-2nd-Ed... [2] http://jeroenjanssens.com/2013/12/07/lean-mean-data-science-...

[ Parent | on HN ]
Re:Data Science Toolbox by paulgb on Wednesday June 04, 2014 @06:26AM (#7844841)
It's a pain to install BLAS and the numpy family on a box you don't have root on so I definitely see value in the VM approach. Does being in a VM degrade the performance to the point where this is mainly a teaching tool or is it comparable to a native install?
[ Parent | on HN ]