Tests say that this is ok

except for

geocoder_tester/world/germany/test_location_bias.csv::Erlange

where now the expected result "Erlangen" comes 2nd and the 1st is "Berlin, Erlanger Straße" which is ok, due to the location bias being exactly there. Reducing the default location_bias_scale to 1.6 would fix this, not sure what we want :)

Then also this would be fixed:

geocoder_tester/world/test_big_cities.py::test_paris_from_new_zeland

Now we have a problem for

/api?q=bahnhof&lat=53.094488&lon=8.806057&location_bias_scale=1.6

and get a 'bahnhof' 40km away from the coordinate as first hit:

{"features":[
{"geometry":{"coordinates":[9.1337314,53.0983492],"type":"Point"},"type":"Feature","properties":{"osm_id":240107774,"osm_type":"N","country":"Germany","osm_key":"place","city":"Ottersberg","osm_value":"suburb","name":"Ottersberg Bahnhof","state":"Lower Saxony"}},
{"geometry":{"coordinates":[8.815475688780303,53.0836703],"type":"Point"},"type":"Feature","properties":{"osm_id":2173872,"osm_type":"R","extent":[8.8148991,53.0841159,8.8163042,53.0832195],"country":"Germany","osm_key":"building","city":"Bremen","street":"Willy-Brandt-Platz","osm_value":"yes","postcode":"28215","name":"Lloyd-Bahnhof","state":"Free and Hanseatic City of Bremen"}}

Update: The best result would be "Bremen Hauptbahnhof". IMO it is a different issue: #318

It would be particularly interesting to know what is a sensible range to choose here

Good question. This depends a bit on the search but probably anything between 0.1 and 10

Hello,

I'm trying to convert a zoom level to a location_bias_scale. Do you have an "ideal" solution? Is the new bbox parameter indirectly doing this?

Thanks!