is there page/link that indicates how “location-names” in the “core-places” dataset are pulled in and adjusted? I’m attempting to tie POI from this dataset with POI from a local POI dataset on name, but it seems many fall out due to simple syntax alterations… It’d be great if I knew those rules so I can adjust and capture more common POIs
Is there page/link that indicates how “location-names” in the “core-places” dataset are pulled in and adjusted?
@Francisco_Avalos_Jr_Data_Angel_at_City_of_Los_Angeles - great question! have you taken a look at the Match service https://docs.safegraph.com/docs/matching-service-overview
this is awesome! I’ll look into this shortly. thanks!
@ross_epstein_safegraph is this a free service?
some of the POI locations from my dataset do not tie out correctly with the POI dataset in the Match service. My POI dataset includes the name of the location, the city, and state. Is there additional information that my local dataset should have to maximize the matching? like lat/long?
Hi @Francisco_Avalos_Jr_Data_Angel_at_City_of_Los_Angeles, matching datasets is one of the trickiest parts of datascience, and like many facets of data science, it is dependent on data. Essentially, the more pieces of the puzzle you can give the match service, the better your odds will be.
Lat/long are not always the best because even with those coords, 2 datasets can have different locations for the same POI (though it doesnt hurt).
Your best bet is to have Name, city, state, zip, address.
How many POI are you working with? Perhaps I can help you come up with a solution!
Hi Jack! Sure, the dataset is not large per se, about 800 web scrapped data points (GitHub - francisco-avalos/LA_covid19_data_cases)
These are the points you are talking about right?
yup, those are the ones I’m talking about
Yeah, so the match service works really well, but without actual addresses, it will be nigh impossible. You might just take the first entry that comes up when you google map query it (API is fastest) and take that address and match that to SG data.
“Skilled nursing” is not a focus here… and I hear that safegraph wouldn’t have information on the “homeless service settings”, but the 3 remaining have about a matching rate of about 70%
hmm ok, i audited those POIs that did match with safegraph and found a few that were outside of la-county, i suppose it’s much better to also have the street address and not just the City, State?
yes very much so, Im not sure I would trust a match without an address or at the very least a zip code attached to it.
gotcha will look into adding these additional infos where possible and view the net benefit
I posted a message here before, but then tried the match service and got the same result. Ill check into this
Hi @Francisco_Avalos_Jr_Data_Angel_at_City_of_Los_Angeles and @Jack_Lindsay_Kraken1 thanks for reporting this; there isnt any limit on the matching so it may be another issue. I will message you both directly to debug further.
It looks like the “match-service” is having glitches? I cannot run it in firefox nor chrome. it just gets stuck in a loading phase
Appreciate you reaching out again. I’ll shoot you a direct message to resolve.
I continue to experience some glitches with no pattern to point to what it is. I notice that my browser freezes when I’m at the point of loading the page (screenshot)… this is true for both chrome and firefox browsers. The one thing I did notice is that the site works normally when I see the chat boc on the lower right pop up, other than that, it just freezes. I’ve also tried login out/in but to no avail