search

forc-db / Global_Productivity

Creative Commons Attribution 4.0 International
2 stars 0 forks source link

Figure 4 (multiple climate variables) #47

Closed teixeirak closed 4 years ago

teixeirak commented 4 years ago
teixeirak commented 4 years ago

Coming back to the issue of rescaled vs multi-panel plots (see issue #44), I'm questioning whether we might not want the multi-panel in the main manuscript. The main reason is that the plotted regression lines are for the rescaled values (I think), and it doesn't really make sense to have separate fits for plotted values vs the "official" fits, which should be on the unscaled data. I think we want to present scaled values for visualization only. Perhaps we could keep the plot this way but rescale the best-fit functions fit on unscaled data? Would that be the same?

@beckybanbury, what do you think?

beckybanbury commented 4 years ago

@teixeirak I'm not entirely clear what you mean - do you mean to keep the graph as multiple fluxes on one plot, but not rescaled? Or do you mean to keep the plot as it is, but change the way we report the values/results?

Scaling the values doesn't change the Rsq/p values or the AIC values (and therefore doesn't change which model is identified as the best fit), but it does change the values of the slopes/regression equations (though I don't think we've reported these).

teixeirak commented 4 years ago

So long as rescaling the values doesn't change the model identified as best fit, it makes sense to leave it as is. My concern would be if the model fit to rescaled values would differ from [the model fit to raw values], rescaled. Does that make sense?

teixeirak commented 4 years ago

This is fine, except for units on temperature seasonality (#68).