Some of you were noticing that the Intersection analysis was a bit slow and we noticed that it was using quite a lot of memory – the larger the dataset, the worse it performed. We tackled this head on and have rolled out a much better algorithm, which, I’ve measured as being between 7 to 10 times quicker! It’s much more zippier now.

The Intersection analysis determines all the locations where one data set overlaps another data set on the map.

For instance you have some areas representing counties and areas representing the pollution plume, that crosses the counties by performing an intersection analysis, you can get new unique areas that represent the bits of counties that are under the pollution plume. It’s pretty powerful.

Or you might want to see all the Coffee Shops that overlap with the “.75 mile buffer of Metro stations” data set. To run this analysis once you have selected “2008 usa starbucks locations” then click “select second data set” and search for metro and add “Dataset from DC Metros Stations with a .75 mile buffer”.

How we do it? We basically only calculate the intersection of the features if they share the same area.

For the geo-techies amongst you, the procedure goes we use goes as follows:

1. create spatial index for dataset 1 features
2. loop over dataset 2 features
3. when current dataset 2 feature intersects with dataset 1 index, then
4. for all the features that are intersected, get the intersection
5. if there’s any errors, try to fix them
6. rinse, repeat