[Geodata] Ok, I'm here . . . (Tiger stuff.)

[Stephen Woodbridge]

Stephen Frost wrote:
> Bob, et al,
> 
> * Bob Basques (bob.b at gritechnologies.com) wrote:
>> Stephen Woodbridge wrote:
>>> Bob Basques wrote:
>>>> I'm using the Tiger data mainly for mapping right now, I'm interested
>>>> in the addressing side as well, but have a good local dataset for
>>>> this, also looking at OpenStreetmap for some comparable value.
>>> I would be interested in this also, but I'm stack overflow at the  
>>> moment so that will have to wait.
>> We have the OpenStreetmap database (whole world) installed in POSTGIS  
>> already, took a few days to just run the index, anyway, it works, and  
>> pretty darn quick, we're not POSTGIS gurus (yet) though, so I'm sure  
>> there is more tuning to be had in the response times
> 
> I believe the OSM data is based on tiger '06se data.  I'm not sure how
> they incorporated it in the end or how hard it would be to update to the
> '07 format though.
> 
>>>> My main interest, and problem area, at the moment with the tiger data
>>>> is the road classification stuff,  It seems overly complicated, but I
>>>> don't know all the subtleties related to who uses what and how.  I've
>>>> been trying to apply the Googlish looking MAP file to it, and it's
>>>> been an uphill battle.
>>> Yes, well I have two major issues in this area.
>>>
>>> 1) CFCC to MTFCC change lost a lot of information about the segments  
>>> as it appears to be a courser gain resolution and it seems to be way  
>>> more complicated. I guess we will have to learn to live with this 
>>> change.
> 
> I'm curious if, overall, information was really lost.  There are alot of
> flags and whatnot and other places where the information could be that's
> not what you're finding in the MTFCC.

Well I can't be 100% sure but I'm pretty sure if you look at the CFCC to 
MTFCC Excel spreadsheet you will see large numbers of individual CFCCs 
collapsed into a single MTFCC.

I can see some reasons for doing this in specific cases, but in general 
I think the data is just getting lost. I'm guessing that the similar 
classification schema was picked because of the effort involved to 
maintain the more detailed scheme, and it can be argued that have a more 
complex scheme that is not accurate is worse that a simpler scheme that 
is more accurate, but that is all speculation on my part.

>>> 2) When you compare Tiger to Google (Navteq or TeleAtlas) the later  
>>> have a major road classification that is not present in the Tiger  
>>> data. These roads are classified as CFCC=A4* in the old data.  
>>> Basically the missing classification in tiger would allow you to  
>>> identify the roads in light yellow on this google map:
> 
> I had a similar issue with the *old* Tiger data, actually, but found
> that they were federal highways and was able to pull the necessary
> information from the federal highway administration shapefiles.

Yeah that might help, but I've worked a lot with Navteq and it just has 
a class of roads identified that are not in Tiger. I actually spend 
quite a bit of time trying to figure out if there was a way to identify 
these roads and mark them, but I did not find any adequate solutions.

>> Yup, this was the first think that I had a problem with.  BTW, I think  
>> Google is using and Average Daily Traffic(ADT)  for their map themesand  
>> not strictly a classification system.  My suspicion was raised when I  
>> noticed some local streets not theming correctly when compared to  
>> Google.  I have no way of know for sure though.
> 
> That's an interesting idea.
> 
>> Do you know of any ADT datasets?   I can't figure out how google would  
>> be able to get this type of info, so that's why it's just a suspicion  
>> right now.
> 
> It could be a dataset that google's picked up from their commercial
> providers...

Yes this is definitely data that is tagged in both TeleAtlas and Navteq.

>>>> I need to come up with a dataset that will extend the coverage area
>>>> beyond our local dataset for centerline prorated geocoding purposes.
>>> So what are you using for a geocoding engine?
>> Some recent extensions to PAGC were developed here under a MetroGIS  
>> contract.   The Engine seems very fast. we're in the testing stages now  
>> with the first extended version.  I believe it will be going public on  
>> the PAGC list fairly soon.
> 
> Sounds interesting.  Public being open source in some way?  I've never
> looked at PAGC, honestly..

Yes, PAGC is open source under GPL, IIRC.
http://www.pagcgeo.org/

-Steve W

> 	Thanks,
> 
> 		Stephen
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> Geodata mailing list
> Geodata at lists.osgeo.org
> http://lists.osgeo.org/mailman/listinfo/geodata