After getting the processing scripts written,[0] I've generated a few sample data files.[1]
This process was a bit unusual because I had two data sets to work with: one set with house numbers and street names on parcel centroids for the city, and one set with house numbers, no streets, on buildings for the whole county. By cross-referencing the two sets of data, I was able to come up with a set of "trustworthy" address points that are corroborated in both.[2] As a side product, a set of "potentially good" points was generated where points could not be corroborated. No tags other than addr:housenumber and addr:street were used. Common errors are: * Points for empty parcels in new subdivisions, where it's obvious that the neighborhood is still being built. Could be outdated imagery too. (guidance appreciated) * Apartment complexes lacking good address points. (don't import) * Mobile home parks with many duplicated points or a single point for the whole park. (don't import) * Long, skinny parcels where the parcel centroid and building point were too far away for the cross-referencing script. (manually review) * New neighborhoods that exist in one set but not the other. (manually review) * Large empty parcels (or farmland) that exist one one set. (don't import) * Random points that are inexplicably missing from one set. (manually review) The sample data was arbitrarily chosen out of the downtown residential area, where it's somewhat less complete than the suburbs. Some of the points can be manually filled in from either original data set. Generally, the suburbs had the best data quality. I can provide more data if people want to take a look. I'm still tweaking little things in the code, but if the data look good, then the next step would be to write a script to split up the data into little chunks and then manually import each chunk. Thanks, Eric [0] https://github.com/erjiang/kokomo-addresses [1] https://github.com/erjiang/kokomo-addresses/tree/master/samples [2] https://github.com/erjiang/kokomo-addresses/blob/master/samples/kokomo_sample_crossreffed.osm -- Eric Jiang, DoubleMap Suite 300W | 748 E. Bates Street | Indianapolis, IN 46202 www.doublemap.com | Office +1(855) 463-6655 _______________________________________________ Imports-us mailing list [email protected] https://lists.openstreetmap.org/listinfo/imports-us
