Oftentimes we have multiple runs of an algorithm on several instances from different groups. First step is generally to reduce approximation sets to scalar value (performance indicators). However, usually we aim for a instance-set based scalarization/aggregation in a second step. Here, one possibility to compare performance of different algorithms among instance groups is via ranking of indicators. I already have code for this in a recent project. Need to wrap this up in a useful ecr function.
Oftentimes we have multiple runs of an algorithm on several instances from different groups. First step is generally to reduce approximation sets to scalar value (performance indicators). However, usually we aim for a instance-set based scalarization/aggregation in a second step. Here, one possibility to compare performance of different algorithms among instance groups is via ranking of indicators. I already have code for this in a recent project. Need to wrap this up in a useful ecr function.