I have been know to get involved with local politics from time to time. About 5 weeks ago, when a federal election was called here in Canada, I decided to build an application that would help local campaign scrutineers get their results in quickly and easily. Election night results and turnout information is notoriously difficult to collect and keep.
An election campaign in Canada is an interesting thing. It is essentially a company that is created, runs for 5 weeks, and then disappears. A cloud solution seemed to be the perfect answer.
I’ll talk more about the application in a future post, but once the application was built, it attracted the interest of one of the major media organizations in Canada, PostMedia. The only thing that they wanted added was mapping.
Another few days of work with the Bing Maps Silverlight control, and we had a couple of fairly interesting apps ready for public consumption. They’re currently available on the PostMedia website here:
The latest version of Reporting Services, due out any day now, has built in support for mapping. This is a very welcome addition and adds significantly to the “cool” factor. I recently completed a quick demo project using public data available from Elections Canada, and I will be sharing some observations over the next few days. I’m a newbie to mapping, so I’m sure that that the GIS folks out there will think this is quaint, but this really does take a technology that was previously available to a select few, and make it available to a much broader audience, something that Microsoft is particularly good at.
Firstly, I have to mention that this was built using the November CTP of SQL Server, and was implemented on SharePoint 2010 beta. Technology Preview on Beta would typically be a recipe for frustration, and while there were bugs, it went together surprisingly well.
There are essentially 2 major data components to a map report,the spatial data and the analytic data. The spatial data contains the map elements themselves,and data that will be used to relate to the analytic data, where the analytic data contains the data that is to be mapped. In my first example, I have data that describes all of the riding boundaries in Canada, along with their metadata (riding number, province, etc). I also have election result data that can be grouped by riding number, so a relationship is created between them. I’ll demonstrate below.
You can create a map in any Reporting Service report, either using the Report Designer in Business Intelligence Design Studio (BIDS), or by using the new Report Builder end user application. Using BIDS, you need to drop a map control onto the design surface to start the map wizard, and Report Builder gives you the option at create time.
The wizard then prompts you for your spatial data. You have 3 choices as shown above. The map gallery lets you choose from the full list of pre-packaged maps available in the product consisting of a grand total of 1 country, the USA. To be fair, this is by design. There are so many border disputes in the world that the US is the only safe bet. In fact, as a Canadian, we share a border with 4 different countries, and unless I’m mistaken, we have border disputes with three of them. This leads us to the next choice, the ESRI Shapefile. These files are the de facto standard for spatial data, and are readily and freely available from the public domain. The Shape files used in this example came originally from Elections Canada. Finally, SQL Server, starting with 2008 supports a spatial data type. This is essentially a series of polygons. It’s a little known feature, and spatial data is supported directly within Management Studio, as can be seen below:
Whenever a query result contains spatial data, a little “Spatial results” tab appears which allows you to visualize the results.
Which one to choose? Well importing the shapefile directly is fast, but using SQL data gives you much more flexibility from a querying standpoint. Also, if the gallery or shapefile are chosen, the data is embedded directly in the report, which is not the case with SQL.
Luckily there’s an excellent little utility available from SharpGIS that makes it very easy to just move your shape files up to SQL.
The next screen you see will depend on your choice, but in the case of SQL you can either choose a shared dataset that has the spatial data, or you can create an embedded dataset for this purpose. Once the data has been defined, you can configure the map’s viewport.
Here you set the zoom level, and can set the panning parameters. You can also choose the type of data that the map is to use – in this case polygon, but other options are point and line (for single point and route type data). Also of note are the two options at the bottom. You can choose to embed map data in the report (removing reliance on the back end data for the map data, effectively creating a snapshot). The second option is to add a Bing Maps layer – allowing you to add the richness of Bing Maps to your visualizations. However, this can only be done if your map data is geographic. Geographic data maps directly to real geographic positions (latitude and longitude) whereas planar data is simply a set of shapes that fit together.
The next window allows you to choose your visualization.
The basic map will only show the data in the map itself, and required no further data. the color analytical map essentially fills in the regions according to analytical data, and the bubble map positions analytical data on the map. Additional layers can be added to any map after the fact. In our example, we choose the Color Analytical Map.
The next screen prompts us to either choose a shared dataset that contains our analytical data, or to create a new dataset. Once we have selected the data to use, the next dialog box allows us to establish the relationship between the map data and the analytical data. In our case, the field FED_NUM in the map data corresponds to the EDANum field in our election results data.
Finally, we choose a theme (which defines the “chrome” of the report), the field to use to determine our our colouring, the patterns to use to visualize the data, and what to display as a label, if anything.
And that’s all there is to creating a basic map! at this point, you can click run in Report Designer, or preview in BIDS, and you should see the results of your analysis as below, which in our case are the results of the Canadian 2008 General Election on a riding by riding basis.
Many things need to be adjusted further (zoom, which color to use for which party, and a number of other things), and I’ll cover some of these in a future blog post, but as you can see, the map wizard is quite capable of delivering immediate results.