[Imports] Postboxes in Czechia, updating details as import

majka majka.zem+osm+imports at gmail.com
Fri Oct 12 11:11:53 UTC 2018


Good afternoon,

as you might have already noticed, the Czech community is actively trying
to get all the existing postboxes on the map.
The data provider for this is the Czech Post. We have a permission to use
these data in writing, the source itself has no licence info included.
In the data, coordinates in local system are included but are neither there
for every postbox nor always correct/exact enough. For this reason, no
direct import is proposed and we are manually checking the location of each
and every postbox. In the data source (a csv file) are currently 20903
postboxes, included is the position for most of them, for all of them there
is an address or location like "supermarket", "bus stop" ... in xxx, and
collection times. At the moment, about half of it the postboxes is located
and is already mapped in OSM. We are writing down the errors and from time
to time sending it back to the data provider, to get it corrected in the
future source files.

*This import should manage the updates needed*, as suggested and as
promised on the talk_cz list.
Some postboxes were removed since they were entered, some are disabled for
an unspecified time and will be enabled again. What has changed the most
are the collection times. Some refs will be changed as well, because of
changes made by the data provider recently. The refs are for this reason
not considered as key in the merge, just the position on the ground.
I have given a very brief description of the proposed import / data update
is in the import log
<https://wiki.openstreetmap.org/wiki/Automated_edits/majkaz_imports/> with
links to the data source and to the scripts and example data. The import
should be done as "majkaz_imports". The source tag would change from month
to month, the format is CP:YYYYMM taken from the month where the data come
from. This is already established as a way to communicate what data exactly
did the last mapper use.

The proposed import will be on very conservative side. The nodes will be
re-tagged only, no positions will be moved, no nodes will be deleted, just
re-tagged as "disused:amenity". The final delete should come a year after
the postbox was last found in the source data, to avoid deleting postboxes,
where for example a construction site nearby is blocking access to the
postbox and it is missing from the data source just for this reason. The
script is checking, if such postbox should be reactivated.

The data source will be separated into individual depots. This will result
in smaller changesets geographically as well as in numbers (biggest group
has about 800 postboxes, average ones have about 300 and there are few very
small ones as well).

To compare the location of existing nodes and the data, I'll be using the
osm_conflator with some hacks to get exactly the result we are expecting (I
am re-introducing the disabled postboxes and removing these again where not
needed to avoid unnecessary changes, and removing potentially new nodes,
for example); as well as a script the Czech community uses to prepare a
geojson file for their instance of poi-importer to mark potential locations
the mappers should be checking.
The scripts used and example files are available on my Github
<https://github.com/majkaz/PBOX_import>, some dependencies to run it might
be a bit non-standard. In the examples subfolder is the complete current
data source and two complete depots included - with all the data, including
the review json file from conflator, the profile and the final osm file for
each. There are only few changes in the example, these two depots were very
well maintained manually.

The import should be repeated monthly, the data file is available from
every 1st workday.
Currently, I have made several tests without uploading the data. I would
like to start uploading as soon as possible, as a real test. The Czech
community is still actively working on it, and will be checking this import
as well.

Let me know if you have any questions or suggestions.

Regards,
Majka
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/imports/attachments/20181012/40792e2d/attachment.html>


More information about the Imports mailing list