[Talk-GB] Tools for using Food Hygiene data e.g. quickly adding addresses/postcodes
gregrs at fastmail.co.uk
Sun Sep 4 19:54:16 UTC 2016
Thanks for your suggestions, Dave F. I’ve created a number of issues in the GitHub repository (see links below). I won’t have time to improve the code for a while, but at least we will be able to keep track of suggestions. I am of course open to pull requests if anyone wants to have a go at improving it.
> The latest total for this district is 1767, so appears a bit more than a week behind.
> Could the update date be added on the page so there's a guide to their accuracy?
There is a date/time at the bottom of each district page, which reflects when that HTML page was generated. The update script gets the FHRS data using their API and downloads OSM data from GeoFabrik before processing the data and creating the HTML pages and JSON files which feed the slippy maps. The whole process takes an hour and a half ish.
The GeoFabrik data is probably a bit behind, but this seems the most convenient way to obtain large amounts of OSM data without having to filter the entire planet file.
> Could a 'Relevant OSM nodes/ways without fhrs:id' statistic be added?
Probably, although the goal of the tool is really to improve the OSM address (and particularly postcode) data, hence the ‘Relevant OSM nodes/ways without fhrs:id or postcode’ statistic instead. (There are generally quite a number of establishments with a postcode but no fhrs:id, although in Bath and NE Somerset I see there are loads of establishments with an fhrs:id but no postcode.)
> A Download (as Geojson?) of the mismatched/missing lists would be very helpful.
JSON files are already created to feed the slippy maps, so could you try [http://gregrs.dev.openstreetmap.org/fhrs/json/overview-270.json <http://gregrs.dev.openstreetmap.org/fhrs/json/overview-270.json>] for example? What were you hoping to use the downloaded files for? Let me know if the linked one already does what you’d like and I can add links to the HTML pages.
> Does it check relations?
No, just nodes and ways at the moment.
> I'm a bit confused. It says:
> 'Relevant OSM nodes/ways without fhrs:id or postcode' = 173
> but then says:
> 'Percentage of relevant OSM nodes/ways with FHRS match or postcode' = 6.3%
> Am I misunderstanding something?
I’ve been through the code that generates the stats again to check what’s going on.
‘Relevant OSM nodes/ways without fhrs:id or postcode’ – seems to be OSM nodes/ways which have neither an fhrs:id that matches one in the FHRS database nor an addr:postcode.
‘Percentage of relevant OSM nodes/ways with FHRS match or postcode’ – this includes OSM nodes/ways which have an fhrs:id and postcode that both match the FHRS database, and those with a postcode but no fhrs:id. This doesn’t include those with a non-matching fhrs:id, or those for which the postcodes don’t agree (including cases where the OSM postcode is not set). This is probably what is misleading because the large number of establishments in Bath with a valid fhrs:id but no postcode will not be included. It’s meant as a measure of ‘completeness’ really, with the goal to tag establishments with a postcode and possibly an fhrs:id. I think the label probably needs changing though.
https://github.com/gregrs-uk/python-fhrs-osm/issues/5 <https://github.com/gregrs-uk/python-fhrs-osm/issues/5> again
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Talk-GB