[Geodata] [Tiger] A few interesting observations on the Tiger2007fe data

Stephen Woodbridge woodbri at swoodbridge.com
Mon Jun 30 00:06:56 EDT 2008


Hi all,

The more I play with this data the stranger it is!

For example, I am finding lots of segments that have multiple zipcodes 
assigned to them. While there is nothing wrong with this, it was 
unexpected and I wonder if this was done based on merge/join with the 
USPS zip+4 data, and when such a merge was last done?

tlid|fullname|paflag|mtfcc|fromhn|tohn|side|zip|plus4|statefp|countyfp
86841497|"Littlefield Rd"|"P"|"S1400"|"2"|"8"|"L"|"01720"|""|"25"|"017"
86841497|"Littlefield Rd"|"P"|"S1400"|"4"|"8"|"L"|"01719"|""|"25"|"017"
86841497|"Littlefield Rd"|"P"|"S1400"|"3"|"5"|"R"|"01719"|""|"25"|"017"
86841497|"Littlefield Rd"|"P"|"S1400"|"1"|"99"|"R"|"01720"|""|"25"|"017"

I also have found MANY records (like 98287 in 1290 counties processed) 
that have multiple house number ranges where the ranges have mixed 
directionality over the segment. IE: on range increases from->to while 
another range decreases from->to along the segment. Some examples (sorry 
don't have the tlid):

WARN: merge_hn(a_from, a_to, b_from, b_to)
WARN: merge_hn(62, 28, 247, 255)
WARN: merge_hn(7, 1, 9, 13)
WARN: merge_hn(98, 50, 55, 57)
WARN: merge_hn(13, 7, 1, 5)
WARN: merge_hn(15, 7, 1, 5)
WARN: merge_hn(600, 100, 602, 606)

I need to change the program that is processing the data to spit out 
more details so I can check the data closer and make sure these reports 
are not symptomatic of a bug.

Some of these look like they might be encoded wrong with respect to 
direction. For example:

WARN: merge_hn(7, 1, 9, 13)

probably is (1-7) and (9-13) which is a continuous range along the 
street segment that was likely split because of different postal codes, 
except that someone swapped the FROM and TO numbers or did not realize 
the segment direction was reversed.

Anyway, more food for thought! Anyone else seen these, have any thoughts 
on it?

Best regards,
   -Steve W


More information about the Geodata mailing list