We’ve noticed for a bunch of the counties in WA, there’s a drop in median home dwell time on Feb. 24. Has anyone else noted the same thing/ any idea why?
Hi @Abby_Smith , I’m trying to replicate the issue tonight to see what’s happening. I’ll keep you posted!
https://safegraphcovid19.slack.com/archives/C0109NPA543/p1589984717003500
@Abby_Smith Tagging @Jessica_Williams-Holt @Marita_Zimmermann from the original post. When I tried to replicate the drop, I got the following graphs (trimmed to show just 2). The graphs correspond to Jefferson County and Mason County. Clearly there’s a drop happening. This make it seem like the drop is happening on Feb 25. Is this similar to what you are seeing?
yep-- weird. We have it on the 24th though?
(So that’s one issue). But also should figure out why that would be…?
I’ll dig into it a bit more. Do you have any graphs or anything else to help me see what it looks like on your side? And just to be safe, we are talking about February 2020, right?
Yes. Thank’s so much Ryan. Here’s an example graph from Jefferson County. Let me know whatever plots would be helpful.
when I filter rows in our data for med_home_dwell <400, here’s some of what I get… Jefferson County is the only one for 2/25 that has < 400. Is there an error in the way we read in the data?
No problem, thanks for the images! One thing that seems strange to me is your ‘med_home_dwell’ column looks more like averages (medians should always end in .0 or .5). How are you calculating that column?
ah! Sorry-- good catch. I need to rename that. It’s actually the weighted mean across census block groups using CBG device_count as the weight.
Okay cool. Even with that, it certainly seems like there’s something weird going on with the data. Here are all the WA counties on the same plot. Maybe there was some external reason for this to happen, but it looks a bit suspicious. I’ll tag @Ryan_Fox_Squire_SafeGraph so he can have a look.
I’m so glad @Ryan_Kruse_MN_State was able to get in on this. i’ve only just got my data flowing correctly. @Ryan_Fox_Squire_SafeGraph i’ve checked some other states and am seeing some similar results. here’s OR
@Abby_Smith just wanted to let you know we are still working on this. It does seem anomalous and maybe is affecting most CBGs, not just the ones in WA counties. Let us know if you figure anything else out or have any other insights.
no idea!
@Ryan_Kruse_MN_State Do you think that this issue could be related to this old issue?
https://safegraphcovid19.slack.com/archives/C0109NPA543/p1585769840329200
There was a known (but unexplained) artifact in the data on 2/25/2020 somehow related to homes that we have not fully resolved. This is Known Issue row 2 on the known issue data tab.
In fact 2/25/20 data is omitted from our dashboard: Not Found | SafeGraph
If we think it could be the same root cause, @Ryan_Kruse_MN_State could you take all of the recent findings and put them into a google doc characterizing the artifact, and I will add it to the documentation of this Known Issue.
If this is a distinct issue, then all the more important for us to document so I can add it to the Known Issues list.
From everything I can tell, this issue seems to have the same root cause as the older issue. I will put together some documentation to add to the Known Issue.
Thanks guys!