data-for-change / anyway-data-science

MIT License
5 stars 20 forks source link

lat and long anomalies #36

Open benadikt opened 4 years ago

benadikt commented 4 years ago

When running the next query:

Select markers.longitude,markers.latitude From involved_markers_hebrew,markers where markers.id = involved_markers_hebrew.accident_id and involved_markers_hebrew.road_segment_number= 80 and involved_markers_hebrew.road1=1 limit 100

all lat and long are pretty much the same except two that are off limit, Write small alg that can find and correct anomalies like that

atalyaalon commented 4 years ago

@benadikt the identifier of an accident is the three: id, provider_code and year you won't find such anomalies when using the three as identifier

atalyaalon commented 4 years ago

Also in both involved_markers_hebrew or makers_hebrew you will find longitude and latitude and road_segment_name or road_segment_number - no need to join them

atalyaalon commented 4 years ago

Just to clarify - not all accidents in the same segments need to have the same lon lat - but you can check me on this one and perhaps prove me wrong?

benadikt commented 4 years ago

not all accidents in the same segments need to have the same lon lat

Lets put them on the map one by one and see the anomalies

Also in both involved_markers_hebrew or makers_hebrew you will find longitude and latitude and road_segment_name or road_segment_number - no need to join them

Also taken but because of platform, and the query as been perform not in real time. there are no significance to join two tables even if not necessity.

https://www.google.com/maps/place/29%C2%B052'50.0%22N+35%C2%B002'41.6%22E/@29.9504694,35.326378,9.5z/data=!4m5!3m4!1s0x0:0x0!8m2!3d29.8805566!4d35.0448942

atalyaalon commented 4 years ago

@benadikt not sure what you mean by that. If you have a specific example of an anomaly please share, if not I think we can close this issue for now.

atalyaalon commented 4 years ago

@benadikt after a discussion wit Dror - Dror will continue exploring this issue - since it's important to validate the road segments we are counting on. @benadikt Great work! @inbalbeka @elashahar01 FYI