# Error estimation in Gaussian data [closed]

I have data set which theoretically is a Gaussian curve . My dataset looks like this:

To find the peak I looked for the maxima in data set. Now this has two parts and off pulse part which is noise, and an on pulse part which is Gaussian. I go for Gaussian fitting because it is not unique, as number of Gaussian fitted have a physical interpretation. My question is say a peak is at $x_{i}$ and what is the probability that given the off-pulse noise of certain variance and mean the peak will shift to $x_{i+1}$?

• It's not so hard to look up things on wikipedia: en.wikipedia.org/wiki/Normal_distribution
– Danu
Jan 26, 2014 at 6:31
• Please edit your question to indicate more precisely how your dataset "depicts" a Gaussian. It would help, too, to know why you are using the maximum to estimate the (location, presumably, of the) peak, given there are superior estimators, such as those described in answers at stats.stackexchange.com/questions/70870 or stats.stackexchange.com/questions/70153.
– whuber
Jan 26, 2014 at 14:32
• @Danu I think the OP actually means there's a functional relationship (with noise) between two variables which looks something like a Gaussian curve, not that there's a sample from a normal distribution. Jan 27, 2014 at 1:33
• @Glen_b Yes! Kind of! I've tried making the question more precise, I hope it helps now. Feb 4, 2014 at 13:21