New sleep start/end times feature is overeager #1277

Closed
by LuccoJ opened 3 years ago · 3 comments
LuccoJ commented 3 years ago (Migrated from github.com)
Owner

screenshot_gadgetbridge_20180921-162959

In short since I'm typing and walking, I think this feature should not count several hours of awake time after a short nap (or in this case, a misdetectrd nap) as just an awakening during sleep, resulting in the idea that I sleep from 17:00.

I think this should be solved by considering only awakenings briefer than about an hour, or half the time of the preceding sleep (whichever is shorter) as being "part of" the sleep session.

![screenshot_gadgetbridge_20180921-162959](https://user-images.githubusercontent.com/1684791/45888532-95d4fe80-bdbe-11e8-88c4-da12f70c7ac1.png) In short since I'm typing and walking, I think this feature should not count several hours of awake time after a short nap (or in this case, a misdetectrd nap) as just an awakening during sleep, resulting in the idea that I sleep from 17:00. I think this should be solved by considering only awakenings briefer than about an hour, or half the time of the preceding sleep (whichever is shorter) as being "part of" the sleep session.
Owner

@elagin Do you have an idea how to fix this? In theory one can have many separate sleep periods during a day. One could take the longest period. A sleep period could still be interrupted by wakeups (short enough to be tolerable). Maybe we need some more structure ini the ActivityAnalysis classes to express this.

@elagin Do you have an idea how to fix this? In theory one can have many separate sleep periods during a day. One could take the longest period. A sleep period could still be interrupted by wakeups (short enough to be tolerable). Maybe we need some more structure ini the ActivityAnalysis classes to express this.
LuccoJ commented 3 years ago (Migrated from github.com)
Poster
Owner

By the way, I'm not sure if it's worth opening a separate issue, but I've thought for a while that the way total slept time is computed should probably be rethought: typically, I look at my asleep time after waking up, but if I've waken up earlier than I did the previous day, it will consider part of the previous day's sleep in the calculation, resulting in an overestimate of the night's slept time. This estimate will then go down later during the day, which is a bit counterintuitive.

So perhaps, aside from the sleep start/end time, the total slept time should only consider the past 12 hours, or if that's not appropriate because it'd miss naps, at least the past 20 hours? Something that makes reasonably sure the previous day's sleep isn't tacked in, at any rate.

By the way, I'm not sure if it's worth opening a separate issue, but I've thought for a while that the way total slept time is computed should probably be rethought: typically, I look at my asleep time after waking up, but if I've waken up earlier than I did the previous day, it will consider part of the previous day's sleep in the calculation, resulting in an overestimate of the night's slept time. This estimate will then go down later during the day, which is a bit counterintuitive. So perhaps, aside from the sleep start/end time, the total slept time should only consider the past 12 hours, or if that's not appropriate because it'd miss naps, at least the past 20 hours? Something that makes reasonably sure the previous day's sleep isn't tacked in, at any rate.
Owner

I believe this has been solved by the Sleep zones, long time ago 371ac276a5

I believe this has been solved by the Sleep zones, long time ago https://codeberg.org/Freeyourgadget/Gadgetbridge/commit/371ac276a5ffa2935faada9a223a751f2f323098 <img src=/attachments/b41d1ccb-fb88-4149-9c31-5f159c66b1b9 width=300/>
278 KiB
vanous closed this issue 1 year ago
Sign in to join this conversation.
No Milestone
No Assignees
3 Participants
Notifications
Due Date

No due date set.

Dependencies

This issue currently doesn't have any dependencies.

Loading…
There is no content yet.