I have not made review of what was imported (primarily because I live on other continent what makes survey tricky, even without COVID) but based on your comments/activity/documentation it seems to be one of the best imports ever done :) Mateusz Konieczny (talk) 21:16, 10 June 2020 (UTC)
Thank you for saying so! I know what I would do differently if I had to do it again, but I'm satisfied with my work. I considered this to be a serious responsibility and I'm thankful for all the community support that I got. This includes your feedback on the wiki page, when I wasn't sure If I was doing the right thing or if anyone was paying attention. I'm just starting to see what I can do for roads in Maine. Too soon to tell.--Blackboxlogic (talk) 16:32, 11 June 2020 (UTC)
- Good luck with that! I guess that roads will require far more manual processing to avoid losing what was mapped already... Mateusz Konieczny (talk) 22:59, 11 June 2020 (UTC)
Dealing with house number clusters
I've just found the E911 import. May I ask, what shall be done with house number chaos like this (Kent, ME)?
- Deletion of duplicates seems the obvious solution. No idea whatever it is some bot bug or duplication in the used source Mateusz Konieczny (talk) 15:49, 13 June 2020 (UTC)
- Short answer: Use your best judgement.--Blackboxlogic (talk) 16:42, 13 June 2020 (UTC)
Long answer: I'm glad you're paying attention. Due to the data source and design descision I made (I'd love to elaborate if you're interested, and I might even if you're not interested), this sort of thing happened occasionally. My options were to import the data as it was, or skip the area. I chose to bring in the addresses believing that it enabled routing, leading a driver very close to their destination, and could be easily improved by future mappers. The map (and addresses) clearly need more attention. I'm hopping that local contributors will do a once-over and reposition addresses, merge addresses with buildings, or delete addresses as appropriate. I believe that I've left things in a state where this is relatively easy.
Regarding "what I shall do": I believe you're specifically referring to the cluster along West Main Street. I see that some addresses hit their target and got merged into buildings, those should be left alone unless you find evidence that they are incorrect. I see some (nine?) addresses were duplicated (I have a guess why*), I would merge those with each other, or delete one. I see many addresses are seemingly IN the road, I would move them to where you think they should be. When there is only one address associated with a building, I suggest to merge the details into the building. When there are multiple addresses in a building, I prefer those as their node sitting within the building near their entrances. This is guidance based on what I would do, please don't interpret these as "rules". Please correct me if you're looking for different or more specific guidance.
- My guess why some addresses were duplicated:
I saw this occasionally and I believe municipalities tracked apartment buildings for regulartory purposes before they tracked every other building. I think they merged these two datasets without handling duplicates. Lewiston had this symptom in higher occurances but I didn't see this symptom frequently accross Maine. My software handled two cases of duplicated addresses; duplicates far apart needed to stay separate otherwise trailer parks turned into a single lump; duplicates stacked close together needed to be merged otherwise apartment buildings turned into a pile. I determined "far apart" to mean 5 meters. I just checked 301 W Maine and they were 7.67 meters apart.
- This is certainly a very good idea!
- What about using some kind of Street View imagery to spot houses with visible numbers? Not for dupes, rather for unraveling house number clusters.
- And last but not least: Are there any more prolific Maine mappers besides emulsi (on osm, edits, contrib, heatmap, chngset com.)?--St Pauls (talk) 21:33, 13 June 2020 (UTC)
StreetView would do it, if you have the patience. I don't, considering diminishing returns and other higher impact tasks. Re: Maine mappers. There are not many mappers in Maine. I'm trying my darndest to organize the ones I find. Please please please sign up for the mailing list https://lists.openstreetmap.org/listinfo/talk-us-maine and/or join the #local-maine channel on slack at https://slack.openstreetmap.us/. We can all pout together! I have literally (virtually) been over every square inch of the state and I hadn't come across emulsi (on osm, edits, contrib, heatmap, chngset com.) or St Pauls (on osm, edits, contrib, heatmap, chngset com.), so maybe there are more hiding? Any time I found a well mapped area I'd check who did it and if they still mapped. I've considered running user statistics on a Maine export, and finding the most active, current users. But then what... send them all a message "HI pLeAse bE My frIEnd"?
People I've found in maine: Sanford! tmerc58 (on osm, edits, contrib, heatmap, chngset com.) has been EXTREMELY industrius in York county. Alan Bragg (on osm, edits, contrib, heatmap, chngset com.) has been working on road names an geometry in Hancock (i think) county, and plans to keep going elsewhere. (He's "from away") Then there's Bkissin (on osm, edits, contrib, heatmap, chngset com.), who I haven't quite figured out.
I'd love to organize a meetup or mapathon or just discuss what we think is important to focus on. Slack is my preference, maybe I'll see you guys there? --Blackboxlogic (talk) 00:19, 14 June 2020 (UTC)