actually, this data is consist of “mass” “cross section” "error"
so I want to draw mass dependent cross section plot with CLs .(x axis:mass, y axis:cross section).
I confused to use roostat.
simple case, I just can draw plot(graph).
but I can not calculate confidence level limit or 1 sigma(2sigma).
I am not sure what exactly you want to do. You have some measurements of cross-sections vs mass with its uncertainties.
CLs is a procedure to find the 95% (or lower confidence value) limit for the cross-section. It make sense to quit a limit when you measure cross-section values consistent with zero. If you have measurements which are significative different than zero, as in your case, you should not plot the limit value but just the measurements and its uncertainties.
You can just use a TGraphError to make that plot
RooStats helps you to find confidence interval values of your cross-section (or limits in case they are close to zero). The confidence intervals are computed from a model description of your experiment. This consists for example from your observed data, the expectations for the single and for the background.