Open source implementation of "feature density" algorithm

Question

I'm working on a dataset of house searches - 1 polygon for each area in which people want to find a house from an online search tool.

This produces many overlapping polygons. It would be great to get a basic understanding of the places that are most frequently searched.

You can do this in QGIS and other programs using transparency:

But ideally, you would create a new layer with the search density for each cell in a grid. This is exactly what the "feature density" algorithm does in ArcView 3.2 and I've seen it work to great effect. It creates a raster map from a mass of overlaid polygons.

So the question is: is there an implementation of this algorithm for my polygon dataset for open source programs. I'm an R and QGIS user so one of those would be preferable. Many thanks!

Answer 1

In GRASS GIS you can achieve the functionality as follows:

• "Point densities are calculated as the number of points per area": See http://grasswiki.osgeo.org/wiki/Count_points_in_polygon (the result is here an absolute result, for density just divide it by area size); or simply use v.vect.stats;
• "Line densities are calculated as the line length per area": Run v.build.polylines to unite line segments properly. Then cut the lines per area with v.overlay (see "Line to polygon clipping"). If categories are meaningful (untested here), use v.to.db to count them and upload to the polygon database table. For density just divide it by area size in a new column (v.db.update);
• "Polygon densities are calculated as the polygon area per area": See http://grasswiki.osgeo.org/wiki/Vector_Overlapping_Areas for an example.

All this can be easily put into a single script. Using the standard GRASS GIS Script parser lines as well, it would auto-generate the GUI, too.