Would this affect the accuracy of the visit counts in {visitor_home_cbgs}, or are those calculated using a separate metric (building footprint, etc.)?

Questions & Help
,
1

Hi all, I’ve been working with the weekly patterns data and found some POI with incorrect addresses and coordinates in the core data - would this affect the accuracy of the visit counts in {visitor_home_cbgs}, or are those calculated using a separate metric (building footprint, etc.)?

2

Examples:
sg:0813c9255a054d9c92679d3bcef027ca
sg:1f87909340874d2481dc3667ff3a8426
sg:2d94d16719844df6b46ad141e6a7255a
sg:37bd369f4eb143edaf74f7704bf9a73f
sg:78ef26b075b9494d864963360b24cd3c
sg:ef1a37690be5431da7c6be1a9e38197e
sg:ef23cce6a7094c2aa46f184e7aa53ea0

3

@Alex_Turicchi_Parkland_Center_for_Clinical_Innovation I am checking the core file now. Which version of core places are you using? Have you read the visit attribution white paper? To my understanding visit attribution is based on polygons, not addresses. But I will double check what you have in core that they are pointing to the right geo-location.

4

@Alex_Turicchi_Parkland_Center_for_Clinical_Innovation So I checked the core file for September and the most recent weekly patterns data, the lat longs do match the addresses but not the names of the location. My inclination is that these locations are mislabelled. Are you focusing on those schools for your research? If so, I am afraid they are not visits to these places but rather, the wrong addresses. Apologies for any inconvenience here.

5

@Ruowei_Yang_UM_Baltimore That was my impression as well, thanks for confirming! We’re focusing on travel to schools in Dallas County, and those were a few of the locations we were hoping to look at in conjunction with metrics from other datasets. We’re currently using travel to a school’s block group (from the Social Distancing metrics) as a proxy for any locations where we don’t have accurate SafeGraph POI travel counts, but to reduce noise the POI travel would definitely be preferred. Do you know if there’s a procedure for reporting these locations to SafeGraph for investigation + correction down the road? Thanks!

6

@Alex_Turicchi_Parkland_Center_for_Clinical_Innovation Is the above a comprehensive list of POIs you found with mismatched locations? I am going to raise the issue to SG staff. Let me know, thanks!

7

@Ruowei_Yang_UM_Baltimore
There were around 60 we identified as mismatches, all are schools in Dallas County
sg:01a48b08c4d445e0a38830e36dd2ba31
sg:0334d27cc59b411b93ee937a74645b9b
sg:0813c9255a054d9c92679d3bcef027ca
sg:170aeb5be6b749329a52932c46a3b941
sg:180218e01af641bbb144f49299b34a85
sg:1ca8286ef08b4d6dbd4f90ef497c83cd
sg:1ee8015565764fb49643609b4307a5b0
sg:1f87909340874d2481dc3667ff3a8426
sg:2073a13ded9c47f4ba4839cc656ef321
sg:2405a87cc62b43f49964f96804b49fef
sg:248085c74013407fab4f54c750bab71b
sg:28c3709e58a74b688bf9eb65b87b9ccf
sg:2d94d16719844df6b46ad141e6a7255a
sg:2ed9e598cd0448d0b17ac3e05698bfec
sg:32945983fee442ab9246a0fda3b6f0cf
sg:32bace16a7f3414a938c1e71c4729d1c
sg:37bd369f4eb143edaf74f7704bf9a73f
sg:37bd369f4eb143edaf74f7704bf9a73f
sg:397d1f5aeca24b51989804e316989160
sg:3a26f0868ba244c4bc84b3627d8a5c8e
sg:3ec48bbe415546eca881f614d725b68c
sg:5a1e1fee746c48e78288235c43f131c3
sg:5b6da8ff0c674b4597434d75ea6f5ce2
sg:682d1879fc7442f3901f841412c1fbe3
sg:7364585a50f442c3b443e909ab5f7696
sg:78ef26b075b9494d864963360b24cd3c
sg:7afa7fa3ae9b43d980fdd8e54b1869c0
sg:8099ac7cf803498682abe964e0e281ac
sg:824174950c204294b4c933edb1f5a629
sg:828f4cd2fa0e45c082e2371722fc8e9d
sg:878c817cb49b45d99f78f7825572737d
sg:8e1b1bc8a72d4806a737dcf275845da2
sg:8e1b1bc8a72d4806a737dcf275845da2
sg:8f5c43d48eca4b96b8dfb02d4f4299f4
sg:99207c797a674632a6981e89653b2109
sg:afb78b3f74304bc9b99096e96cddf147
sg:b3487fbc254a471d840bbd46d7fc5e05
sg:b4303c6e42c04c87a899dbfbd997d8e5
sg:bbbcf68b7a4540118e522897f18de2c1
sg:bf7907a742904731aa99c2f32c6c1c5f
sg:bf7907a742904731aa99c2f32c6c1c5f
sg:bfceeaaffd9241baa29231f3bcf40389
sg:bfceeaaffd9241baa29231f3bcf40389
sg:c0b1c37ecc3b486fb942c1ea3e012151
sg:cbacb4119e354719acd1529c70b317ec
sg:d6608227e6c24fc3a3fa10a0db4e7b20
sg:d92e9711e8c04fc9ad3d7675c6e123ad
sg:e2ae822547f740579d493e353f6e38b9
sg:e9876900c6884af290b22e3a338a9031
sg:ed9f4aaa58fa4b23a4ace6b3581bc911
sg:ef1a37690be5431da7c6be1a9e38197e
sg:ef1a37690be5431da7c6be1a9e38197e
sg:ef1a37690be5431da7c6be1a9e38197e
sg:ef1a37690be5431da7c6be1a9e38197e
sg:ef23cce6a7094c2aa46f184e7aa53ea0
sg:f2210e4f16a94fdaaf7a5560f7021525
sg:f250644bebf3419daf87abaf790337d1
sg:fd1710fc826241a2820f7596f18935ac
sg:6bb59f66c1444b51ba4fb45a2fed8768

8

Thank you for reporting this, much appreciated! @Ruowei_Yang_UM_Baltimore

9

@Alex_Turicchi_Parkland_Center_for_Clinical_Innovation Sure, this is important and thank you for compiling the list.

10

@Alex_Turicchi_Parkland_Center_for_Clinical_Innovation Do you have the correct addresses for those schools? I am assuming you are familiar with the Dallas area in general. Schools data are a work in progress and the product team is working generally to fix some related issues. But if you can provide a 4-column CSV with the correct addresses in the format below, they can specifically target those POIs.
• safegraph_place_id
• field_name
• orig_field_value
• fixed_field_value
The field_name here would be street_address and you can input the original (wrong address) and fixed (corrected address) values.