Sunday, April 29, 2012

Fun with Fusion Tables

I've been playing around with Google Docs features a lot in my spare time. For instance, playing with the different document types (Documents, Presentations, Spreadsheets, Forms, Drawings), as well as Google Apps Scripts. One tool I hadn't done much with, but had heard some about, are Google Fusion Tables. At Google I/O 2011 there was a presentation I sat through on Fusion Tables, and although it looked really cool, I got the impression that it was best with large amounts of data to aggregate through - not really what I was using Google Docs for at the time.

A walkthrough of some Google Fusion Tables features.

Fast forward to today, and I've been using Google Docs a lot more. I've collected Wii Fit metrics to help lose weight. I went to a Google Hackathon in D.C. and discovered some of the conveniences of Google Apps Scripts. So the other day when I saw Tables available in my Google Docs, I figured I'd finally try it out.

Finding a Data Set

I took part in the AFCEA 5k in the Baltimore/Washington area, but was underwhelmed with the text file of results that they provided on their site. So I did what any rational person with a computer and a bit of free time would do: I dropped it into a Spreadsheet (in this case, courtesy of Google Docs). Let me take a moment to mention that the importing the data provided through AFCEA was less than perfect - some participants excluded hometown information or age data, and others didn't finish the race (probably because they lost their timing chips). So importing isn't super smart yet, but after moderate Google Spreadsheet editing I persevered.

Now this Spreadsheet has information for 762 participants. If you're used to Google Spreadsheets, you may know that working with a somewhat sizable data set sometimes isn't so great - sorting and searching data is sometimes slow and a bit of a nuisance. The most interesting part of this data however, was that it included hometown information. In Spreadsheets you can't do much with that, but I remembered from the demo last year that plotting data to a map was easy with Fusion Tables.

Initial Impressions of Fusion Tables

Fusion Tables are a little hidden under the Create button of Google Docs.

From there, you get a friendly little prompt asking you what data set you'd care to import from. Options include spreadsheets and text files, Google Spreadsheets, or creating an empty table. I imported from the Google Spreadsheet, although I found editing batches of data from Fusion Tables was somewhat tricky (and also a feature request). So I got the data established after re-importing the same data multiple times, and imported in this Table format. If you view the data set, you'll notice a few things: it renders fast (I assume because of there's no Spreadsheets bloat of loading full/large amounts of data or Apps Scripts), it sorts fast, and it has a few other different options.

One new option is to merge data sets, which is good if you're combining data from multiple sources. Another is to import more rows, which should be good if you want to work with very large amounts of fixed data - something I've found inconvenient through Spreadsheets.

The Neat Stuff

Filters and Aggregators are another tool that's new, and by far one of the more interesting tools. The AFCEA data had a few typos in it, and although I caught most of them in Spreadsheets, there were a few imported into the Table. Finding the discrepancies by filtering down was really slick, as I could filter for the typo in a particular field, and the matches came up instantly. Aggregation was also easy in Tables. For example, the graph below displays the time (in seconds) it took the max, min and average runner to complete the 5k. It's a lot slower a process to represent this data in Spreadsheets.

What really was neat in Tables though is that you can also map the data, and you don't need coordinates to go it. Fusion Tables can derive locations from record data and "geocode" it, giving you a cool way of presenting table data. A little spooky that it's so easy to process, but impressive. Also I hear there's a limit to the amount of data you can freely geocode (like 2500 records/day), but that's pretty decent for personal use.


So there's some neat things that Fusion Tables have to offer. I'm still toying around with this, but so far it's been pretty awesome to work with. If you're looking to compare this tool to something in the Office suite, your closest bet is Access, but with data sharing courtesy of the Internet, and visualization tools courtesy of Google gadgets that make the experience 500 times better. So think of Fusion Tables as Access, but much, much more awesome.