MSG 3100 5.3
Bias
Average of your Error. (The sum of all errors/ the number of errors.) If positive means under-Forecast If negative it means over-Forecast If Zero we cannot assume it to be perfect model because there can be large negative and positive error.
Mean Absolute Deviation
MAD Absolute Value of your Error. (Sum of absolute value of errors/Total number of errors) Takes absolute value of error terms which eliminate problems of bias.
Mean Absolute Percent Error
MAPE (Absolute Value of Error/ Absolute value of sales)
Mean Squared Deviation
MSE Error Squared. Square absolute value of errors. (Squaring of error/Total number of error.) Squared error is used so that large error gets more attention. By squaring large error gets increase by square value.
Root Mean Squared Error
RMSE The square root of Mean Standard Error sqrt(MSE)
Error
The difference between what actually happened and what your forecast predicted. (Actual - Forecast) can be positive or negative