It seems like the end of the year sees a regular pattern of drops .

This is amazing, maps.me has been the most user friendly OSM app i’ve used by far. Going to try editing tomorrow.

One of the very interesting observations in Indian cities were they all evolved from little villages or grew around a fort a large temple. The patterns of streets and buildings are organic and radiate along directions that are the shortest path to the next village or settlement.

This is very different from the gridded cities of US, where the convenience of the planner took precedence over the plan or aspirations of the residents.

MapMakinMeyers: The query for the raw data is in the post: http://overpass-turbo.eu/s/f5N

Vincent de Phily: Thank you. The boundaries on OSM was updated using various sources, should have clarified that.

Yes, we don’t have a consistent addressing scheme or street signages in India. Finding a home usually involves asking for a landmark like an important junction, building or shop and then asking the local people for the exact house.

This is an amazing exercise. We have really poor footpath infrastructure in India, especially in terms of accessibility for visually impaired and those on a wheelchair, although I suspect Kerala is far ahead on these matters.

Every community has their own approach to tagging footpaths as separate ways. I’m for mapping them separate especially in town centers with heavy foot traffic, this allows for much more precise pedestrian routing. But on residential roads, it makes more sense to have it as an added attribute sidewalk on the road.

It seems like the common understanding of a halt is a minor station with limited services on the railway line. With this definition, it certainly does not make sense to tag subway stations as halts.

Would there be a better way to tag the type of a station, especially terminal stations and interchanges which one would want to be more visible on a map?

Awesome! Wish there was a MapCSS setting to not display small features as POIs so everyone can see the same style.

Aha, thats a hidden setting.

Settings>Map>Don’t display small features as POIs.

Don’t think data based styling is supported :?

Beautifully documented joost. This is a great tutorial in mapmaking in itself.

Wonderful analysis! Running this for the whole world will be quite revealing about the state of data on the map.

Btw, this is the Kannada map of the world http://yogiks.github.io/osm-kn/map/

Very cool! A spreadsheet is definitely the most convenient interface for translators to use for such a task. You should also check with the Kannada OSM folks how they translated over 35000 objects which is the highest for an Indic language on the map.

This is going to be really really useful for us in India. Thank you so much.

Using bank address information is quite interesting, since this information is available fairly openly. The map is looking amazing in the area, nice work!

@philippec not everyone has the leisure time to dedicate effort to learning OSM and mapping, and this is where experienced contributors like us can help bridge the gap.

Here is the wiki mailing list https://lists.wikimedia.org/pipermail/wikimedia-in-wb/

Also worth a try is the wiki village pump https://bn.wikipedia.org/wiki/উইকিপিডিয়া:আলোচনাসভা

Might be useful to reach out to the local linux user group or Bengali Wikimedia community.

This is such a great application of OSM data! Have on quite a few occasions been in situations where I wanted to quickly find an ATM on the way, but had no way to do so, till now :)

Tried it out and works great for my area. It would be nice if I had the option to control the buffer size and limit the radius of search results. Could you share the link to the repository? Might try doing this over the weekend.

@andygol thats true ;)

Also JOSM used to add all my loaded layers before, even if it was invisible. This was patched recently to include only visible imagery layers.

Thanks Yogi, this is useful. Atleast we now know the numbers are not completely off. One thing to note is that both the pl and the osm-coverage script will be giving slightly higher counts due to dual carriageways. Still uncertain how badly this affects the final value.