Dear all
I am trying to visualize poor access to broadband services in Denmark and have data that cover all residential addresses in Denmark. That is, for each known address, the data shows if there is access to broadband and which speed is currently offered to consumers plus the ets89 coordinates with 2 decimal point precision. The data consist of roughly 2.4 million addresses of which about 300,000 thousand have access to speeds of 100 mbit/second or less (incl. 20,000 with no access). For convenience I classify these as poor access. I am using Stata 16 IC.

I am able to compute a proportion of poor access by municipality and turn this into a choropleth map using spmap. And I can also overlay each address coordinate as a point on the base map using the following syntax:

spmap using C:\..\edges.dta, id(stid) ocolor(gray) fcolor("253 157 36") point(data("C:\...\points.dta") xcoord(koord_X2) ycoord(koord_Y2) size(*0.01) fcolor("63 126 68"))

The problem with that visualization is that the size of the marker is big and therefore densely populated areas will appear less well served because points cluster here. And it doesn't change if I alter the scaling factor to, say, size(*0.0001). If I adjust the size of the marker manually in the graph window, and also change its opacity, I can sort of get Stata to comply. However, it seems that saving the edited graph requires a bit too much memory for Stata, and I am unable to save the resulting graph. The un-edited graph is 25 megabytes when saved as a pdf. I have uploaded a png of the un-edited map for access to 50 mbit/sec. or less so you can see what sort of crazyness I am up to. The end result I am looking for is less green and more orange by making the markers smaller. I also want to avoid green color in the ocean because markers transcend the edges of the polygons.

Does anyone have an idea of a good hack here? (Apart from just doing a choropleth map and give up this silly venture.)
Array