[talk-au] QLD GTFS Data Imports
jasonjward5 at gmail.com
Sun Feb 23 16:14:12 UTC 2014
I've been sporadically looking at the best way forward on this import and
am at a bit of a loose end. I am still to document the process for
importing but in doing so and monitoring the @imports list for a while now
one of their significant considerations is how best to maintain the data
after import. As noted earlier there was an import done in 2009 by user
morb_au of approximately 6000 bus stop nodes and with 13,000 nodes now
available from the agency I think I now understand where the OSM
contributors are coming from with regards to mechanical imports.
I was originally inclined to simply delete these in favour of the 'newer' /
'more authorative' dataset in its place but the inclusion of the
qroti:surveyed value  in the existing dataset has given me pause to
consider the accuracy of what is there and whether it is better to leave
some or all of it in place and conflate with the newer, more extensive
dataset from Transport and Main Roads.
To understand what the surveyed value means I have looked back through the
Archive version of the qroti site, I've checked morb_au's diary notes ,
 on the import and even read morb_au's paper on SVG . I have sent
morb_au an OSM message and even tried hooking up with him on Linked In with
me at 0/5 so far. My only requirement being to understand where the
qroti:surveyed is from (a government employee survey date or a morb_au
My thinking is that morb_au didn't survey 6,000 bus stop across Brisbane
and that this was information held in a dataset that he had access to.
The extension of my thinking, and this is where I'm hoping someone can
check my logic, was to retain nodes in OSM that are version 2 or higher and
delete or merge version 1 nodes with the gtfs key:value information from
the newer dataset. There are 4,641 v1 nodes with the qroti:surveyed value
in them. my thinking here is that if stops are at v1 then they are as
imported by morb_au and those nodes are a mechanical import (not supported
by extra information or enhanced by another more recent OSM mapper edit)
and candidates for removal and replacement with the nodes in the newer
Is that a good approach?
While I'm asking questions would it be prudent to deprecate (read: remove)
the qroti values of nodes as they are merged? They hold no value as they
are sourced from an import whose source is indeterminate and the values of
it  are no longer useful given the currency of the TMR dataset I am
trying to get done. Any thoughts on that?
 : http://taginfo.openstreetmap.org/keys/qroti%3Asurveyed
 : http://www.openstreetmap.org/user/morb_au/diary/5499
 : http://www.openstreetmap.org/user/morb_au/diary/5421
 : http://www.svgopen.org/2003/papers/PublicTransportMappingExample/
 : http://taginfo.openstreetmap.org/search?q=qroti#keys
 : http://mapcraft.nanodesu.ru/pie/368 (5kx5k Mapcraft Pie)
On 26 December 2013 16:03, Jason Ward <jasonjward5 at gmail.com> wrote:
> Hi everyone,
> I took the discussion offline momentarily and Daniel and I have had a few
> email exchanges. The result is a Wiki page with information about the
> process. Anyone interested in helping with the import and QA process (both
> documenting and doing) is more than welcome to go right ahead.
> With 13,000 rows in the SEQ stops.txt file and roughly 9,000 existing
> nodes in the region the work of checking will need to be segmented with
> http://mapcraft.nanodesu.ru/ being noted by Daniel as a useful tool (I
> never knew it existed - thanks mate) to help with that endeavour.
> Anyway, the page is up and you'll note a lot of TBA in sections. If you
> know what should be done in those sections please chip in and update the
> documentation (or just delete it).
>>>> Let's start a write up and take it to the imports list for further
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Talk-au