Hi All — — @James_Custer_UT_Austin and I are wondering two things.
- What is the difference the between “median_home_dwell_time” and “median_percentage_time_home” ? And,
- What is the difference between “candidate_device_count” versus “device_count”?
Thanks!
hi @Paul_Rathouz_UT_Austin
did you read the documentation where these columns are defined?
If you are still confused, can you be more specific about your confusion? happy to help!
So regarding “device_count” vs “candidate_device_count”, I am gathering that the former excludes devices without any activity in that time window. Correct? If so, what is “activity”? Thanks! looking again at the other two now.
Regarding “median_home_dwell_time”, I am gathering this is the total number of minutes the device is detected at home during the time window. For “median_percentage_time_home”, I would have thought that is just the same thing, multiplied by (100/(24*60)). But, there is the clause “time we observed devices at home versus observed at all…” in the documentation. So, is the denominator different then 1440 minutes here?
@Paul_Rathouz_UT_Austin sorry for the delayed response.
You are correct about the difference between device_count
and candidate_device_count
regarding dwell times
No, we do not observe devices every minute of every day.
A useful mental model is that we are getting a psuedorandom sampling of moments throughout the day for a device (“pings”).
Please see this discussion for more context.
https://safegraphcovid19.slack.com/archives/C0109NPA543/p1585778381343300
the percentage_time_home should be essentially time_home / (time_home + time_not_home)
But that denominator will almost never be 24 hours, because we are only able to receive data for a sampling of teh day.
Please feel free to @ mention me if you have any additional questions