[Talk-transit] [Talk-gb-westmidlands] NaPTAN and the new PTtagging schema

Peter Miller peter.miller at itoworld.com
Fri Jun 26 18:23:51 BST 2009


On 26 Jun 2009, at 17:51, Andy Robinson (blackadder-lists) wrote:

> Peter Miller wrote:
>> Sent: 26 June 2009 4:41 PM
>> To: Thomas Wood
>> Cc: Talk-gb-westmidlands at openstreetmap.org; talk-transit at openstreetmap.org
>> Subject: Re: [Talk-gb-westmidlands] [Talk-transit] NaPTAN and the new
>> PTtagging schema
>>
>>
>> Your suggestions below make a lot of sense. I would however very much
>> encourage you to include customary stops because they do indeed
>> 'exist' even though there is no physical pole. Consider a road that
>> doesn't have a name plate but when you people who live on the street
>> what it is called they tell you. Does the street have a name or does
>> it not - I suggest we would agree that it does? If a tree falls in a
>> wood and there is no one to hear it did it make a sound etc.  
>> Customary
>> stops can be confirmed by looking for physical marks of vehicles
>> stopping or people standing around on the grass, from information at
>> the stop opposite or from asking bus drivers. I would suggest that  
>> for
>> now we believe NaPTAN.
>
> These are easy to add in a final cleanup anyway, just by usage of  
> the route.
> The problem with the NaPTan data is that there are loads of stops  
> that are
> probably just not used at all, hence we leave them turned off  
> (silent data).
> I agree that we could and probably should import customary stops but  
> I don't
> think we should assume they are actual in-use stops and hence should  
> leave
> them silent in the database until someone confirms and adds  
> highway=bus_stop
>
> For other areas of the country I think its fine (with the exception  
> of CUS
> stops) to go ahead straight away and add the highway=bus_stop where  
> there
> are few existing mapped stops. Ideally a post to the local uses in  
> the area
> would confirm either way what they would like to do.

You seem to be putting out different messages in the two above  
paragraphs. Are you saying you support the import of CUS stops or not.  
Also are you suggesting that bus stops are set as 'real' (ie active)  
stops.

Possibly Roger will have some views on how many unused stops there are  
likely to be in the dataset. Looking at the Oct08 dataset there were  
365,000 bus stops and 42,020 of them were unused at the time however  
this doesn't necessarily mean that they don't exist, only that no  
buses currently use them - in some cases they could be stops for  
summer-only services. I suggest that we should include all bus stops  
in the dataset regardless of use. We should removed stops that don't  
physically exist if there is no sign of them on the ground. Customary  
stops might need a visit to the friendly local bus operator who  
probably has all the information in his head. Physically marked stops  
can be checked by cruising the bus routes.

>
> Beyond that the only bit of data I dislike from the original run is  
> the
> unverified=yes tag. It would be better to change this to verified=no  
> for
> future imports (and easy to swap in West Mids.)

sounds good
>
> Otherwise my experience in Brum is generally good in that with the  
> exception
> of location (which is 10m to 100m off at least 50% of the time) the  
> NaPTAN
> data matches the data on the ground very well.
>
The accuracy will vary across the county and will reflect the care  
taken by each authority. I would expect it to be better in most places  
but might be proved wrong!

Having a map that shows the bus stops would seem to be a good step to  
getting it improved by doing a physical survey or asking bus drivers  
to comment. If the data is hidden in the maps and not exposed it will  
be harder to sort out. I vote for having the data introduced as fully  
visisbly data but possibly we do it county by county. I am happy to be  
an early recipient of data for Suffolk and I think Ed Loach is keen to  
see the Essex data.



Regards,



Peter


> I know Brian and others have documented a few oddities here:
> http://wiki.openstreetmap.org/wiki/NaPTAN_Error_Log
>
>
> Cheers
>
> Andy
>
>
>
>
>
>
>>
>> Traveline would strongly advocate for their inclusion so that OSM
>> links seamlessly to their journey planners.
>>
>>
>> Regards,
>>
>>
>>
>> Peter
>>
>>
>>
>>
>> On 26 Jun 2009, at 16:21, Thomas Wood wrote:
>>
>>> 2009/6/24 Peter Miller <peter.miller at itoworld.com>:
>>>>
>>>> On 24 Jun 2009, at 18:20, Thomas Wood wrote:
>>>>
>>>>> 2009/6/24 Peter Miller <peter.miller at itoworld.com>:
>>>>>>
>>>>>> Can I suggest that we treat this import and any final tagging  
>>>>>> as a
>>>>>> separate
>>>>>> issue on separate timeline from the NaPTAN import just so long as
>>>>>> no
>>>>>> important information in the NaPTAN DB is lost in the process.
>>>>>
>>>>> Can you clarify what you meant by this?
>>>>> Is it essentially that we don't care about the new tagging schema
>>>>> and
>>>>> get on with the import?
>>>>
>>>
>>>> Yes. I would suggest that to avoid trying to agree a new tagging
>>>> arrangement
>>>> in a hurry prior to the import and keep the two projects separate.
>>>> Firstly
>>>> we import the rest of NaPTAN as agreed in the original discussion,
>>>> and then
>>>> secondly we agree a harmonised tagging arrangement of some sort and
>>>> convert
>>>> all the data to this new format (including the NaPTAN import).
>>>>
>>>> btw, did you mean this to be off-list? Feel free to copy the thread
>>>> to the
>>>> list if it was a mistake.
>>>>
>>>>
>>>>
>>>> Regards,
>>>>
>>>>
>>>> Peter
>>>
>>> Ok, then to get on with the import, we need to review the errors we
>>> made with the Birmingham trail, and to get their views on the data
>>> review process - was it a good idea to import things without the
>>> highway=bus_stop tag, to get people to add them themselves?
>>>
>>> I think the one other outstanding issue is how we should represent  
>>> the
>>> CUS stop types, at present in the 'active' tagging mode, they'll
>>> appear as fully-fledged highway=bus_stop nodes, like every other bus
>>> stop type, but with the addition of  naptan:BusStopType=CUS, as (a
>>> rather obscure) indicator to the fact they may not exist.
>>>
>>> And then finally, we need to think about how we roll this out,  
>>> county
>>> at a time is the most obvious step, I think we order the import  
>>> based
>>> on requests on the transit list, followed by requests on talk-gb,  
>>> with
>>> a target date to import the rest by.
>>>
>>> And on the technical front, I'm going to have to make sure that the
>>> import tools I'm using are 0.6-capable.
>>>
>>> I'm copying this over to the west-mids list so we can get their
>>> responses.
>>>
>>> --
>>> Regards,
>>> Thomas Wood
>>> (Edgemaster)
>>
>>
>> _______________________________________________
>> Talk-gb-westmidlands mailing list
>> Talk-gb-westmidlands at openstreetmap.org
>> http://lists.openstreetmap.org/listinfo/talk-gb-westmidlands
>





More information about the Talk-transit mailing list