Filtering fluctating data
I have a few simulation where the values I'm interested in don't completely converge but fluctuate around a certain value.
While the monitoring points are fluctuating the error values get continually smaller.
First set of data, seems easier:
First set of data, seems harder:
Since I never took any proper statistics class, I have no clue on how to proceed here.
Is it correct to say: if the error is smaller than 1e-4, the solutions is acceptable. Therefore I average all the monitor values that have a correlating error smaller than 1e-4 - even if the monitor values have clearly not converged at the first few iterations that meet the criterium of having an error smaller than 1e-4?
Or is it better to say: I can see that my monitoring values are converging, so no matter what the error of the simulation is, I just average all the monitoring values that seem to fluctuate around a certain value rather than showing a certain trend?
Hard to describe my problem, I hope you got what I mean ;)
|All times are GMT -4. The time now is 23:08.|