This paper proposes a novel procedure based on a benchmarking model of data envelopment analysis (DEA) to solve the two-group classification problem. The idea was inspired by the fact that benchmarks dominate their own production possibility set (PPS). If
