Https://root-forum.cern.ch/t/unable-to-display-the-histogram-entries-and-standard-deviation/50280

[gini]

Hi,

I am trying to make comparison of some already published data with my results. For which I need to extract data (Gaussian shaped histograms mainly). So what I do is I extract the X and Y data points from a graph and I re-plot using the method discussed in " Unable to display the histogram entries and standard deviation". When I do this I really wonder how the errors are calculated by root say, for the mean and standard deviation.

As for understanding better, I have histogram from a ttree which I am attaching below:

So i extracted the data points and I replotted the data using the method described in “Unable to display the histogram entries and standard deviation”. Please see below :

As you can see from the plots the error obatined when I replotted the data is quite different from the true error which the data has.

Can anyone tell me how to resolve this problem. How can I obtain the correct errors when I replot the data point.

Any insights about the error calculation in root would be really helpful.

Thanking you for the kind consideration!

[yus]

Hi Deby,

I think your best option is to extract the errors, too, just like you extracted the X and Y values.

[gini]

But the thing is not always that I have errors properly evident from the distribution. For example :

So how do I resolve this problem then?

[yus]

Maybe you could contact the owners of the plot? The errors could have been set manually and do not follow any evident logic. Trying to guess the errors can be a bit unreliable.