Interval or limit based on the (strong) assumption of Normality. Means and SDs for, if you are trying to place a confidence Whatever looks Normal, *that* is what you need to find That it will not be the "linear power" represented by dB. What is it that looks normal? dB? log(dB)? I'm nearly sure When you take log(dB-measures), you now have a plot that The basic data are dB, which is already the log of power.īut the assertion seems to me as if you *may* be saying that What do you have to do to your dB measures to see Googling for several hours has not yielded any concrete results. >I'd appreciate if anyone who could shed some light on this issue. > when checked with the normplot() function. > For what it's worth, the data is Gaussian in log scale (10*log10) I'd appreciate if anyone who could shed some light on this issue. I doubt I can directly apply std() to the dB data either.ĭo I need to implement the standard deviation function manually, using x and x_bar in dB? Is there a matlab function for this?įor what it's worth, the data is Gaussian in log scale (10*log10) when checked with the normplot() function. It boggles my mind as to why I can't do this. What about the std() function though? If I convert the dB values to linear, compute the standard deviation, then bring it back to dB, the answer is outright wrong. So, I would convert to base 10, take the mean, then convert the linear mean to a dB value. It is my understanding that I cannot directly use the mean() function as the decibel values need to be converted to base 10 first.
So the idea is to compute the mean, add two standard deviations, get a result in dB, and use that in my subsequent calculations. The purpose is to characterize the back lobes from many measurements and account for 96% of cases (2 standard deviations above the mean). I'd like to get the average and standard deviation of this data between several patterns. I have a bunch of power measurements (from antennas) in dB. I'm trying to do some statistics on data in MATLAB, but this is more of a fundamental mathematics question than anything else.