Sleepyhead Statistics

[bradleyjoe]

Why in sleepyhead the last 7 days AHI off. I look at daily add up the last 7 days and divide by 7 and the number is not the same. Is their a problem or how does sleepyhead calculate it. Brad

[ChicagoGranny]

Why in sleepyhead the last 7 days AHI off. I look at daily add up the last 7 days and divide by 7 and the number is not the same. Is their a problem or how does sleepyhead calculate it. Brad

Calculate the weighted average - multiply each AHI by the number of hours of usage for that night. Now take the sum of those numbers and divide by the total hours for the seven days.

That is the weighted average and should be equal to the SH number. Never tried it myself, but let us know if SH agrees with your calculation. It should.

[palerider]

Why in sleepyhead the last 7 days AHI off. I look at daily add up the last 7 days and divide by 7 and the number is not the same. Is their a problem or how does sleepyhead calculate it. Brad

Calculate the weighted average - multiply each AHI by the number of hours of usage for that night. Now take the sum of those numbers and divide by the total hours for the seven days. Now divide that number by the total hours for the seven days.

That is the weighted average and should be equal to the SH number. Never tried it myself, but let us know if SH agrees with your calculation. It should.

my brain hurts!

[Nick Danger]

I've found that Sleepyhead has a more accurate approach to calculating averages than ResScan.

[ChicagoGranny]

my brain hurts!

Well no wonder! I duplicated the last instruction. Corrected now.

Take two aspirin and call me when you are back PailRiding again.

[ChicagoGranny]

I've found that Sleepyhead has a more accurate approach to calculating averages than ResScan.

Did you ever actually calculate some weighted averages and compare them to SH and ResScan?

[palerider]

I've found that Sleepyhead has a more accurate approach to calculating averages than ResScan.

mark tends to obsess about those kinds of things.

[Nick Danger]

I've found that Sleepyhead has a more accurate approach to calculating averages than ResScan.

Did you ever actually calculate some weighted averages and compare them to SH and ResScan?

Yes. ResScan tends to do a weighted average of the medians and the various percentiles. Mark actually loads the detailed data for all the nights and calculates them directly. You can't legitimately average medians or percentiles (even if you weight them correctly).

[palerider]

my brain hurts!

Well no wonder! I duplicated the last instruction. Corrected now.
Take two aspirin and call me when you are back PailRiding again.

thank you granny, much better now. your concern warms my heart.

[ChicagoGranny]

ResScan tends to do a weighted average of the medians and the various percentiles.

That's strange. It's harder and takes more lines of code to do it that way.

[Nick Danger]

Well, the advantage is that they can report averages without keeping the detailed data. The display on the machine for various time periods can't be done using detailed data - because they don't store that much. I'm guessing that they don't want their software to report different numbers than the machine itself.