Removing all roads that don't have associated addresses removes roughly 35'000 which would seem to be much more compatible with the previous data, so I'll switch to doing that.

Am 12. November 2019 09:22:49 MEZ schrieb Simon Poole <simon@poole.ch>:

Am 11.11.2019 um 13:29 schrieb Simon Poole:
....
My current suspicion is that the difference is due to streets and places
that don't actually have postal addresses associated with them now being
included, I couldn't pin this down with any local data, anybody else
have a theory?

....

I've found a good example of this in Birmensdorf (ZH) that used to be
near 100% coverage and now has 49 street matches "missing".

Explanation: the attributes for streets and places contain a value that
indicates if the object is actually used in the GWR (as in: there was an
address associated with it), that value has now been changed from "not
used" to "used" for streets that don't actually have an address
associated with them, or just have a building without an actual postal
address.

The former means there is no actual geometry associated with the street
in the GWR making it "difficult" to locate the road. I could however
conceivably filter these out based on the fact that there are no GWR
addresses with the street name, the later is more difficult to resolve
as the address data distributed by the BfS doesn't contain any of the,
in principle available, meta-data on building type and so on.

Any opinions on how to proceed?

Simon



--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit Kaiten Mail gesendet.