[OSM-talk] Activity statistics per city (or region)

Mateusz Konieczny matkoniecz at gmail.com
Wed Oct 21 21:45:43 UTC 2015


On Wed, 21 Oct 2015 17:00:49 +0200
César Martínez Izquierdo <cesar.izq at gmail.com> wrote:

> Hi, we are interested on getting some OSM statistics per city.
> 
> The idea is to get a rough idea of the activity of the community on
> each city, by counting the number of active users or the number of
> existing nodes on the city (which could be then divided with the total
> number of inhabitants in order to get a relative indicator).
> 
> As we only need rough estimates, we could also use region statistics
> if they are easier to compute or already available. Our area of study
> is cities having >= 50000 inhabitants in Europe.
> 
> Do you know if someone has already produced such (or similar)
> statistics? If not, do you have some ideas on how to compute them?

Counting nodes is a poor idea. For start - only recently changed nodes
should be counted (edited during last three month?). There are many
cases of imported data that just sits there and slowly gets outdated
as nobody maintains it.


Counting number of active users may work better. It will suffer less
from problems with imports - also, just because there is one massively
active mapper does not mean that there is "community".

But it may be tricky to exclude remote mapping - for example many
places in Nepal were edited by many people but does not mean that there
is any local community. But in Europe it may be a lesser problem.


I would also consider using number of solved notes (within some
time, like last two months). Notes may be less affected by imports
(though it may be skewed by note spam).


Probably some composite of multiple factors would give the best result.



More information about the talk mailing list