At Wisconsin, we wanted to find a way of estimating and providing a cross check for the live time of the AMANDA-II detector for 20 percent of the 2000 data. Since each file contained approximately 700 seconds of data, we divided the year into 55 minute bins and plotted it.
Plot of livetime in 55 minute bins.
The projection of this plot should (and does) give us a Poisson distrubution about the average number of events per bin.
Projection plot of the 55 minute bins.
To calculate the live time, we then added up all the bins that had greater than zero events in them. Then we divided by 5 (since we are looking at every 5th file). This gives us 44.5 days of live time.
Taking into account a duty factor of 80% we find the live time to be 35.5 days. If the duty factor was 70%, the live time would have been 31.1 days.
My live time estimation: 31.1 to 35.5 days. This is consistant with the number found by Mathieu Ribordy at Zeuthen.