Coda on shop completion rates on OSM

Thanks to John Baker (Rovastar) for a few suggestions discussing my recent blog post in the pub last night:

• What do the graphs of numbers of unique shop tags look like with heavier filtering of relatively poorly used tags.
• E-cigarette shops are a recent phenomenon, and should represent a genuinely novel tag rather than the mix of typos, synonyms etc which characterise much of the long tail of shop tags.

These were easy to follow up, so I present the graphs here:

Unique shop tags over time on OpenStreetMap for Great Britain,
filtered to remove tags with a restricted number of uses as at June 2017.

For virtually any level of filtering the curves level out around 2010-2011. Thus the core set of shop tags looks to be very stable. A good place to judge the extent of likely synonymy for shops in Britain is the LUA script used by SomeoneElse for his "Useful Maps".

Growth of mapped e-cigarette shops in GB on OSM

As expected e-cigarette shops first appeared rather late, at the end of 2013, and there are a decent number mapped (over 200 by mid 2017). I haven't checked, but I suspect the sharp increase in 2017 was caused by some tagging rationalisation. It's not unusal for new things to acquire a range of synonyms before tagging stabilises and one value becomes favoured. (It's equally true that in some cases this does not happen).

I've had a couple of other requests which it will take rather longer to look at, but if you have ideas relating to shops in Great Britain I can look at the data right now.