If a user has two different query sets, e.g. titles and people, RunExperiments currently reports the rollups for each set and the rollups as if there were not sets (for everything). It would be useful to allow users to weight the query sets for a combined score...I want to see how well I'm doing with a composite score where we weight the titles queries performance score by 80% and the people queries by 20%.
This could also be used in the GA for the target evaluation metric...not just a rollup of all queries, but a user-chosen weighted average.
If a user has two different query sets, e.g. titles and people, RunExperiments currently reports the rollups for each set and the rollups as if there were not sets (for everything). It would be useful to allow users to weight the query sets for a combined score...I want to see how well I'm doing with a composite score where we weight the titles queries performance score by 80% and the people queries by 20%.
This could also be used in the GA for the target evaluation metric...not just a rollup of all queries, but a user-chosen weighted average.