Hi all. I am new to SafeGraph. I have a question about the accuracy of POIs. Did any folks check whether a POI’s location name matches its geographic location and street address? I explored POIs in Yellowstone National Park and extracted 80 POIs from the SafeGraph dataset. However, after checking POIs by Google Map, there are only 40 POIs that match their location names, geographic locations, and street addresses.
Hi @Yun_Liang_Penn_State_University, I will give some information and you let me know if it can explain what you are seeing.
1st, it is important to note that SafeGraph does not take data from Google Places, but rather scrapes multiple sources to create the dataset we work with (similar to how Google works), thus there are bound to be discrepancies in remote locations like Yellow Stone – in terms of which is more correct, I typically will try to search the location organically and see if I can find a website that confirms either of the addresses (Google or SafeGraph). If that doesn’t work, you can always try checking against a 3rd party API, just make sure it isnt just a portal for Google Places
In terms of geographic location, if you are referring to Lat and Lng, how far off are the points? lat/lng typically vary quite a bit from place to place which is part of what makes Placekey so useful
If you would like, post 5 or so POI from the 40 that don’t match and we can try and work through which is which.
Thank you for this @Yun_Liang_Penn_State_University I will dig into this tomorrow and let you know what I find!
Hi @Yun_Liang_Penn_State_University , just letting you know I am a bit behind but havent forgotten about this! I will get back with my findings asap
No problem. Take your time.
@Jack_Lindsay_Kraken1, I have a question about how your company derived the POI information? Can I help with SafeGraph technical team to improve the accuracy of POIs’ information? There are 400+ national parks in the US and correcting those POIs could be an interesting project. My advisor and I hope we can get accurate POIs for this project. Thank you again!
@Yun_Liang_Penn_State_University i will ping @ross_epstein_safegraph on this one. He will know how best to proceed. Whatever the outcome may be, thank you for the generous offer!
hi @Yang_Pan_LSU, thanks for going through these. We are working on a correction intake tool - basically a formatted google sheet we could give you that runs a script to submit feedback to us. It would be great if we could try to collect your feedback through that. @Jenny_Doan_Safegraph can provide the details.
Hi @Yang_Pan_LSU, yes we are testing the feedback tool internally and once it’s ready for further testing, it would be great to get input to ensure it captures the requirements. I will keep you updated. Thank you!
@Yun_Liang_Penn_State_University I believe you were meant to be tagged in the messages above. Just wanted to make sure you saw these.
@Jack_Lindsay_Kraken1 Hope you are doing well. I was wondering if you and your team check the accuracy of POIs in Yellowstone National Park. I looked through the documents in SafeGraph about how did SafeGraph derive its POI database. But I am still confused about how did SafeGraph derive POIs, such as natural attractions, in national parks. This type of POIs is different from restaurants, offices, or other locations since natural attractions cover large areas. How did SafeGraph determine their accurate latitudes and longitudes?
@Jack_Lindsay_Kraken1 Another question is if I use my phone without cell access, will SafeGraph (mobile applications) still capture my data? Thank you.
@Jack_Lindsay_Kraken1 For some national parks, like Great Smoky Mountains National Park, have their own POIs to represent the entire parks, while some national parks, like Yellowstone National Park, do not have their own POIs. I am interested in potential reasons to cause this issue.
@Ryan_Kruse_MN_State @Thomas_Young_Econometric_Studios_Utah_Legislature do you have any insights on this?
@Spencer_Vail_SafeGraph I think they just need to be grouped that way (POIs within a broader national park), but haven’t been. But, I would check with data engineering. On the cell service: My understanding is that geolocation gets tracked even when cell service is down, so I think location is tracked in this case, but I haven’t found a definite answer from an authoritative source. Might check with engineering.
@Spencer_Vail_SafeGraph Hi Spencer, I was wondering if we could set up a meeting with SafeGraph technical team. We do have many questions that need to be answered thus make SafeGraph data validated. This is important for SafeGraph as a company too since we are trying to validate their data for national park use. Thank you so much!
Hi @Yun_Liang_Penn_State_University, we have office hours coming up on Wednesday 5/26 from 3:00-3:45 CST. Could you share a few details about the questions you have so we can prepare a bit beforehand?
@Ryan_Kruse_MN_State Thank you so much! My research is about using SafeGraph data to explore visitation numbers and visitor demographics in national parks. First, I identified 80 POIs in Yellowstone National Park and checked POIs’ addresses and latitudes & longitudes in Google Maps. However, only 40 POIs’ addresses match their latitudes & longitudes. I noticed that other people also reported this issue. Secondly, I am interested in how SafeGraph derives POIs, such as natural attractions, in national parks. This type of POIs is different from restaurants, offices, or other locations since natural attractions cover large areas. How did SafeGraph determine their accurate latitudes and longitudes? Thirdly, for some national parks, like Great Smoky Mountains National Park, have their own POIs to represent the entire parks, while some national parks, like Yellowstone National Park, do not have their own POIs. I am interested in potential reasons to cause this issue. The last question Another question is if I use my phone without cell access, will SafeGraph (mobile applications) still capture my data? Thank you.