I’ve added a Loss.py file, which calculates the Pierson correlation and the p-value of rejection of correlation for the LogLoss and the QuadraticLoss of one “true” random model that is global and 10000 randomly drawn models. I got (0.90794121110503834, 0.0) when I ran it, where 0.908 is the correlation and the p-value of rejection that the two measures are correlated, p =0.0.
I’ve added a Loss.py file, which calculates the Pierson correlation and the p-value of rejection of correlation for the LogLoss and the QuadraticLoss of one “true” random model that is global and 10000 randomly drawn models. I got (0.90794121110503834, 0.0) when I ran it, where 0.908 is the correlation and the p-value of rejection that the two measures are correlated, p =0.0.