Expressing L5 and M5 timing and how this affects averaging across days

[vincentvanhees]

L5 and M5 timing in part 5 are converted timing in hours relative to previous midnight. When averaging across days this makes sense for variables relating to the timing of sleep, but not necessarily for the timing of L5 and M5.

For example, 13 (1pm) and 35 (11am), will average to become 24 (midnight), which is not informative because the real average is noon.

If we change the numeric timing to be the hour in the day, like we do in part 2 we will have a different problem: For example, 11pm and 1am will result in an average timing of noon, because (23+1)/2=12, while it should be midnight.

So, either way averaging the timing of an event across individuals is tricky if we do this based on numeric timestamps. Would be good to see how other packages solve this issue.

[vincentvanhees]

It seems there is no solution for this, except that we may need to update documentation to warn about this.

[vincentvanhees]

Closing this issue now, I have updated the part 5 output variable documentation in the vignette now. To warn user about this.