Dear ROOTers,
I have a question about statistics.
What should bin error be when bin content is zero or less than one number that we cannot think of it as Gaussian distribution?
I know the bin error is equal to 1/sqrt(N) when bin content is under normal distribution.
So what is bin error when bin content is equal to 0?
Hi,
the absolute error equals sqrt(bin content) or sqrt(sum of squares of weights). So when a bin content is zero, the absolute error on this bin is also sqrt(0)=0.
OK , what should bin error be when the bin content is so small that it don’t obey the normal distribution?
So what is the bin error now?
Thank you for your reply~
I don’t think bin error has anything to do with normal distribution. If we fill a histogram with weights = 1 and if bin content = 2, the error will be sqrt(2)/2 = 0.71. If bin content is 1, the error is sqrt(1)/1=1. If the bin is not filled, the error is 0.
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.