Good that Google is offering courses for people with no knowledge of stats, journalist should be prepared to handle all the open data that is being released. Reminds me of the case of Irene Choge a journalist who attended a Google sponsored Bootcamp and later discovered flaws in certain schools in Kenya. source: http://blogs.hbr.org/2013/03/open-data-has-little-value-if/
There already are a pair of data-driven journalists at the WSJ whose day-to-day sounds more like a data scientist than a typical journalist. They are Tom McGinty and Rob Barry.
They gave a great talk recently, which I wrote about (shameless plug below), summarizing how they investigated the Asiana airline crash in SF recently and what their day looks like. They're brilliant:
This seems pretty cool, I'll definitely give it a go. Great to see Google offering this, and it looks like a much smaller commitment (10-15hrs) than many MOOCs.
I'm interested in which data experts created their "What is the primary goal you hope to achieve by signing up for this course?" radio options on registration though, they don't seem mutually exclusive. :)
In the data-cleaning portion of the course, I hope they make use of OpenRefine, which was formerly Google Refine, when Google bought it as GridWorks...years later, it's still the best highly-specific, GUI-driven data tool I've ever used, such that I'll gladly hop out of my normal programmatic-workflow to scout datasets with:
I'm interested in the portion teaching Fusion Tables, as the official documentation is still catching up to the recent redesign and API changes. I put together a basic guide for a class I taught last fall, but relied on my vague experience with the older version of FT and was unaware of any new features they may have added: http://www.smalldatajournalism.com/projects/one-offs/mapping...
It just looks like an elaborate way to market FusionTables. But it's a popular move for everybody these days, and even a couple of MOOC's have been used to spread the word about some professors pet-project/platform.
But putting that aside, I don't see why they didn't support any of the other free/open platforms (like R). (I'm expecting "but it's too complicated to that user-segment" argument)
The data processing concepts that will be purportedly taught are not tied to any platform or language. I've used Fusion Tables to teach concepts of data and it's mostly useful for the built-in mapping capability and the ability to merge tables via foreign key, and the pivot/summary transformations. It's not just that beginners lack the tools to do these in SQL/Access/R/etc., it's that they don't even know what these concepts are, or what the value in them are.
If you don't know many data concepts, then this course is probably worth taking. Then apply them to your datasets wherever they may be stored
R, Pandas (python) and alike will do a good job IMHO. I've been working with Matlab before (not really statistics, but more general), and now R is really delightful with visualizations and quick analysis.
If you're looking for a rich programming language with basic analysis function I would recommend Pandas. Not as diverse as R, but works pretty good. (Just watch out for that steep learning curve in the beginning).
