Tag Info

New answers tagged

0

The CartoDB administrative regions geocoder doesn't include county data at the moment, so you'd need to upload a county dataset and merge it with your data table. You can use the merge option in CartoDB, or you can use a SQL query in order to copy the geometries, such as: UPDATE data_table SET the_geom = (SELECT the_geom FROM counties_table WHERE ...


0

Chris W is on the mark, especially on the final two points. A hundred block is of limited valuable and a street name without a number at all is only useful if there are just one or two blocks to the street. As he indicated, treat each "class" separately, though the first two likely can be run together depending on the locator style. E.g., a US Streets style ...


1

The numbered street records are standard and not a problem assuming a match can be found in the geocoder/source data. Depending on the service, accuracy should be very high. Some are based on actual point or parcel locations. Others simply use percentage length of the line and the range for that line to establish where along the line to put the point. Cross ...


1

ISO3 and ISO2 codes should work always for the countries. I have done a test with the four regions that you added and all of them seem to be working correctly for me -- you can check the table here. In order to geocode it, I used the "administrative regions" option, with "description" in the first selector, leaving the second one empty. Do you have any ...


1

Pelias gets boundaries from external source (or form OSM but loaded via separate query) and join points with boundaries within ElasticSearch. Try this https://github.com/kiselev-dv/gazetteer/tree/develop/Gazetteer It uses pretty similar scheme (with some additions) as Pelias. Run gazetteer.jar from releases with: bzcat CH.osm.bz | java -jar gazetteer.jar ...


0

The city names geocoder that CartoDB offers uses data from GeoNames, which is using "New York City" for New York. This is why by using New York the one from the UK was appearing, as it's selected due to the amount of population. We have done some changes with respect to the synonyms of the cities and now geocoding "New York" will give you the city in the ...


0

I updated @scw's script which scrapes the data from Wikipedia. It now uses requests instead of urllib2, Beautiful Soup 4 and it outputs a JSON instead of writing to a CSV file. import json import bs4 import requests print(json.dumps( [ { ['name', 'alpha_2', 'alpha_3', 'numeric'][no]: td.find_all()[-1].text ...


1

Looks like the USPS has a tool now! I just found out about it today. It also seems to be free. https://www.usps.com/business/web-tools-apis/address-information.htm


0

I'm going to post this answer but hopefully another contributor will help break down the following, which I think will paint a more coherent picture: What is the impact of the number of states loaded on geocoding? I've got all 50 and I'm seeing a much lower performance than @LR1234567 (i.e., 8x time per geocode). What's the most efficient method for ...


1

Geocoding is generally unavailable for synced datasets in CartoDB, because each sync would cost geocoding credits which would end up charging the users an exorbitant amount. However, if you have columns called "latitude" AND "longitude", "city" or "IP addresses" your georeferencing will be done automatically upon each sync. Maybe that could be a workaround ...


1

try wikimapia. you can search by rural route number, city name, postal code etc. If someone has put any of this info on wikimapia, you can get coordinate from there. you can import wikimapia info into google earth by a simple utility tool (ge.kml) which you can download for free from www.wikimapia.org/ge.kml. Open ge.kml in google earth where you can pan ...


0

Perhaps http://openplaques.org. You can obtain address, coordinates and details on the cleanliness of the original geocoding for many of the plaques. Not a very even geographical or language distribution, and also slightly odd addresses in general, but it's an interesting dataset to use.



Top 50 recent answers are included