Review: Online databases let you structure and share your data

Google Fusion Tables

This is not an industrial-strength database -- for one thing, Google still labels it as "an experimental application from Google Research." While it's been around for years and Google employees have offered assistance to journalists posting public Fusion Tables projects, it nevertheless doesn't scream "Trust my mission-critical application here."

Fusion Tables isn't meant to be a SaaS version of a database platform like MySQL. Instead, it is specifically designed to not only store data in tables and easily merge them by common keys or columns, but to "gather, visualize and share data tables" -- with the emphasis on visualization. For developers who want to do more with Fusion Tables data, there is an API that includes SQL select, insert, update and delete queries. The graphical user interface does include filtering capabilities as well.

Besides merging, sorting and filtering, Fusion Tables offers relatively easy ways to create charts, maps and network graphs; it also lets you easily search for public data (this is Google, after all) to merge with your own data set(s). And Fusion Tables' most compelling feature is indeed its data-visualization capabilities, particularly the ease of turning tables of data into color-coded maps.

To take the map-making capabilities for a spin, I uploaded data by state from the U.S. Census Bureau's 2013 American Community Survey on how many households have a computer and broadband Internet access: table B28003 at the easy-to-use Census Reporter website.

After uploading my spreadsheet (Excel, CSV and Google Sheets as well as KML geographic data are all accepted), Fusion Tables immediately recognized the state names as geographic entities and started off with a basic map of the data -- by basic, I mean it showed points for each state, all the same color. It was relatively easy to color-code the point icons by selecting "change feature styles."

However, to turn this into a color-coded choropleth map where polygons are shaded based on data values, I needed to give Fusion Tables the necessary geographic-boundary files. The standard for such files is typically shapefiles, a format developed by Esri, the creator of ArcGIS. Most geographic-boundary files from the US Census Bureau are shapefiles, for example.

However, Fusion Tables uses (Keyhole Markup Language (KML)) files. The Census Bureau has started releasing some KML files; otherwise, there's a free shapefile-to-KML Web converter specifically for Fusion Tables, or you can search for an already-existing public Fusion Table with the geography you need.

In my case, state geography is a pretty common need, so I searched public Fusion Tables and found an existing KML file that I copied into my own Google account.

Next, I needed to make sure there was a common column in both tables referring to states by the same name -- using full state name, two-letter abbreviation, geoID, etc. In this case, the "state" column in my data table and "name" column in the KML file both use full state names. I joined the tables by selecting File > Merge, in my data table, selecting my geography table and then choosing the state and name columns respectively. I chose the columns I wanted and created a new merged table that contained both my data and information about state geography.

In that table, the "map of geometry" that was created automatically showed all the states shaded in the same color (red). I chose "Change feature styles" under "Feature map," selected "Fill color" under Polygons and chose "Gradient" so the colors would go from light (low number) to dark (high number). Clicking "Show a gradient" allowed me to select my data column -- in this case, the percent of households with a computer and broadband Internet. I took out some columns in the pop-up window by selecting "Change info window" and fairly easily created the map below.

Note that files imported into Fusion Tables can be a maximum of 250MB and each user gets 1GB of storage.