“…Some papers have been published on the subject with ATCOs (Borack, 1995) but with the democratization and improvement of machine learning, a new interest could emerge for these algorithms. Inspite of this, a simple method can be better than complex ones (Nusinovici et al, 2020) and explicability should be developed: a "black-box" is difficult to certify (Tippins, Oswald, & McPhail, 2021). Based on Pareto-optimality concept (De Corte, Lievens, & Sackett, 2007), elaboration of composite scores and their weightings may be improved in order to consider validity, adverse impact and constraints (De Corte, Sackett, & Lievens, 2010).…”