I am really struggling to find a way to create statistics on OSM data by myself.
Aiming to review the impact of certain events on the OSM contributions, I want to retrieve statistics of the amount of edits within a certain time frame and a certain region.
These statistics should be on an hourly or daily interval and don’t have to be live. Even if I have a delay of creating the data a week after the actual edit, it would be OK.
I now started to do some reading, but mostly due to my lack of understanding databases, OSM files and stuff, I am failing to find my way through.
I am aware of the weekly released planet dumps, but I doubt these can help me.
Much more interesting might be the changeset dumps, but these seem to store only the different edit sessions, which can include from 1 to 20000 edits and might thus be misleading.
How would you tackle this problem and do you think someone who has quite limited coding and database skills can manage it?
Best regards from Cologne
OK, I digged a little bit deeper and it seems to me, that this might get even more complicated than I imagined.
Currently, my last hope lays on the Planet dumps combined with the daily/hourly diffs http://planet.openstreetmap.org/replication/day/000/000/.
If I get it right, I could go back to an old, weekly Planet dump and then update the data step by step with the diff files until I cover the time frame that I want to examine.
After that, I will use Osmosis to define the region that I want to focus on.
Do you think this might work or do I get the function of these diffs completely wrong?
It is not easy. I´ve been working on a similar project for some time now. I think you´re best of working on the history files. There is a world dump available, but I would definitly recommend downloading smaller files, as available here http://download.geofabrik.de/
There is a great tool to cut those up for a bounding box or a polygon (in poly format), this turoial is really helpfuil
https://github.com/MaZderMind/osm-history-renderer/blob/master/TUTORIAL.md . If, like me, you have no experience with Linux, it will take you some time te set things up.
I wrote a diary entry about my experience here https://www.openstreetmap.org/user/joost%20schouppe/diary/21826
For what you want to do, I think you might want to query edited nodes-ways-polygons within a certain timeline and a certain area. Easiest is to only import this area into your psql database, then dump a text file out of it. Do get in touch if you want to get some more details.