Skip to content

Ranking Modes Explained

Mode Behavior Use Case
M1 Ranks strictly by Metric 1. Single-metric evaluation.
M1_M2 Metric 1 is primary. Metric 2 can correct (swap) ranks if a lower-ranked method significantly outperforms a higher-ranked one in Metric 2. Balancing performance vs. cost.
M1_M3A Metric 1 is primary. Metric 2 acts strictly as a tie-breaker. Tie-breaking without overriding primary results.
M1_M2_M3 Full hierarchy. M1 is primary. M2 corrects M1 (iterative). M3 applies two sub-logics: (1) One-time correction if M2 is neutral, and (2) Iterative tie-breaking if both M1 and M2 are neutral. Complex multi-objective benchmarking.