I am wondering how should I understand the numerical value of the correlation score for each updates. For instance, for each day, we will get the updates for the numerical correlation value for each rounds. As my model is still developing, my model for each rounds is always different…
As we usually have 4 values (one for each rounds) for every day updates, I am wondering, so, is the correlation is dealing with same Meta-Model? Or different one?
For instance, my round 1 have a lower correlation but higher MMC, my round 2 have a higher correlation but lower MMC. I am thinking to ensemble them together by assigning some weights, but are they dealing with same model or not when taking about the correlation?