Closed skyinthecloud closed 10 years ago
After a quick search, I was relieved to find that no community areas nor census tracts so far appear to "share" any of the property centroids as in the issue above. How I know:
select lda.pin, count(*)
from landbank_data_censustract as ldc, landbank_data_assessor as lda
where ST_Intersects(lda.loc, ldc.loc)
group by lda.pin
having count(*)>1;
and
select lda.pin, count(*)
from landbank_data_communityareas as ldca, landbank_data_assessor as lda
where ST_Intersects(lda.loc, ldca.geom)
group by lda.pin
having count(*)>1;
both yield zero rows; therefore, no PINs from the assessor file are associated with more than one tract or community area. This is good news.
Also, ST_Intersects will return true even if there's only a line intersection, i.e. a parcel is right on the border. So this is not a problem.
Anyone care to figure this one out? In the assessor data, there's exactly one property that, by way of spatial query against our Wards geometry data, falls in BOTH of two adjacent wards. Run this query in postgres to see what I mean:
Which for every other PIN would yield exactly one ward number, but for the ward above gives:
I see two possibilities: the point for this parcel is on the boundary of two coterminous wards, or wards 7 and 10 overlap on this point.
Anyways, it's 1 point in ~2 million, so I'm ignoring it for now. As it is, the property is "vacant land" according to the assessor data.