The seek of high precision analyses in $\gamma$-ray astronomy leads to the implementation of multivariate combination, benefiting from several reconstruction methods. Such analysis, called $X_{eff}$, was developed for the H.E.S.S. data using three shower reconstruction methods. This paper presents the improvement granted to this analysis by refining the distribution calculation of discriminant variables, considering observation conditions, and adding new variables in the $X_{eff}$ combination. The efficiency of the analysis is presented using simulations and real data. A comparison with the standard analysis (model++), for a typical set of sources, shows a significant gain in sensitivity.
↧