benchmark is accuracy or loss?

closest-git / QuantumForest

Fast Differentiable Forest lib with the advantages of both decision trees and neural networks

77 stars 19 forks source link

benchmark is accuracy or loss? #5

Open showkeyjar opened 3 years ago

showkeyjar commented 3 years ago

It's strange that the best accuracy is smallest number.

something wrong?

closest-git commented 3 years ago

Sorry for the misleading. Some testing results are copied from NODE, which is actually some distance between the predictions and the targets. So in all other experiments, I'm also using the same distance to compare the accuracy. This could be confirmed in the code.

showkeyjar commented 3 years ago

ok, what is NODE or node distance?