An important part of our work in T-16 is devoted to provide the laboratory and the international nuclear physics community with very precise nuclear data evaluations, to be used as basic ingredients in numerous applications (e.g., nuclear reactors, particle accelerators, astrophysics, nuclear medicine, etc). Such evaluations are usually the result of various physical models, continuously improved, along with experimental data sets, especially important when the available nuclear models are known to fail.
While experimental results are unavoidably blurred by uncertainties, both statistical and systematic, a Bayesian inference scheme can help determine a "best" estimate for the physical quantities considered.
I will present a Bayesian study we have recently conducted on the neutron-induced fission cross-section of U-235 and Pu-239. A particular emphasis will be put on the practical aspects of the evaluation work, especially the crucial phase of data analysis. Possible improvements over the current tools we are using will finally be introduced (proposed?).