fable
fable copied to clipboard
Request for different opt_crit Values
I am wondering if it is possible to add new error metrics to optimize forecasts on. RMSLE is a powerful metric for forecasts that should get penalized less for missing high or by high %.
RMSLE is defined quite well here: https://hrngok.github.io/posts/metrics/#:~:text=Root%20Mean%20Squared%20Logaritmic%20Error%20(RMSLE)&text=It%20is%20the%20Root%20Mean,possible%200%20(zero)%20values.