Computer scienceData scienceMachine learningRegressionRegression metrics

MAPE and sMAPE

Approaching infinity

Report a typo

MAPE is known to suffer from the fact that, in certain cases, we might get arbitrary big scores. What can we do to prevent this?

Select one option from the list

FInd the maximum possible score and re-normalize all the obtained scores

Divide MAPE by 2

Cap MAPE at a certain threshold (like 100)

Remove the absolute values from the terms

___

Create a free account to access the full topic