For the Vector -> Research Tools -> Random Point -> Stratified sampling design, can we have:
Add "Proportional Value From Input Field" selection, where the user can select a floating point field, and then they could enter either: 1) Number of points on map, or 2) Value of each point. If the user selects number of points, the algorithm would sum the positive numbers in all the polygons, and sum the absolute value of the negative numbers, and divide the max of these two sums by the number of points specified, to get the value-per-point. (E.g. if the user specified 1000 points, it would yield no more than 1000 points representing positive values, and no more than 1000 points representing negative values.) Then the algorithm would iterate through the polygons and for each polygon would divide the absolute value of the user specified field by the value-per-point, round it to the nearest integer, and place that many random points in the polygon. Each point would be labeled by the value of the field divided by the number of points. Advantages over current Random Point -> Use value from Input Field: 1) Don't have to generate a new field representing the number of points, if you have a field representing something big and continuous, like Gross Domestic Product per state. 2) Don't have to figure out the value of each point in advance, can specify that you want 1000 points (for positive values) and get 1000 points generated. 3) Generates negative values as well as positive values (I often use blue dots for increases and red dots for decreases in dot-density maps.) 4) Random points are assigned a quantity such that the original value can be retrieved with a sum (i.e. although the number of dots is necessarily rounded to an integer, the attribute on the dot values can vary between the different polygons.) (Unless zero points are generated in the polygon.) 5) (related) works on floating point fields. PS I have this working well in PostGIS and have used it for years. Function attached. It's just annoying to use if the data aren't in PostGIS already, and I'd like to rest of the QGIS community to have this cool functionality. I always thought I'd implement this myself in QGIS code, but I'm starting to realize that I'm neck-deep in too many coding projects right now. However I would buy beer or pay someone to implement this, if it helps speed it along. -- John Abraham
AddDotDensity.sql
Description: Binary data
_______________________________________________ Qgis-developer mailing list Qgis-developer@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/qgis-developer
