Currently the response variable is the mean of the log NPP in june, july, and august of the final year of the simulation. It could just as easily be the log of the mean NPP and these are not mathematically equivalent. I'm not entirely sure which is more appropriate if the goal is to back-transform the predictions (or if we are even interested in back-transformation). I should think about this a little.
Currently the response variable is the mean of the log NPP in june, july, and august of the final year of the simulation. It could just as easily be the log of the mean NPP and these are not mathematically equivalent. I'm not entirely sure which is more appropriate if the goal is to back-transform the predictions (or if we are even interested in back-transformation). I should think about this a little.