Given the fact that all measures of accuracy suffer to a certain extent
from advantages and disadvantages, two accuracy measures are used:
mean absolute percentage error (MAPE) and root mean square percentage
error (RMSPE).MAPE underscores the magnitude of the error more
than the direction of the error. RMSPE penalizes larger errorsmore than
small ones and hence is more appropriate when the variable of interest
exhibits fluctuations and turning points (Witt &Witt, 1991). Both forecast
error statistics are scale invariant and thus permit comparison
across different time series and for different time intervals. The MAPE
and RMSPE are defined as