How You Measure Forecast Accuracy

You analyze the overall performance of a plan by measuring the forecast accuracy. You can use mean absolute percentage error (MAPE), mean absolute deviation (MAD), and forecast bias to measure the forecast accuracy.

MAPE

MAPE is a measurement of the percentage difference between the actual value (actual shipments) and the statistical forecast. The higher the MAPE value, the larger the difference, which demonstrates a high degree of forecast error. MAPE calculations are the central basis of forecast error analysis and you can use it for statistical, loaded, and manually entered forecast measures.

Predefined measures to display MAPE are the following:

• Final Bookings Forecast 3 Month MAPE

• Final Shipments Forecast 3 Month MAPE

Bias

Bias is an indicator that supplements MAPE and describes whether the demand is typically higher or lower than the forecast. The Bias function calculates the percent difference between two measures. When the Bias value is positive the demand is greater than the forecast. When the Bias value is negative, then the demand is lower than the forecast.

Predefined measures to display Bias are the following:

• Final Bookings Forecast 3 Month Bias

• Final Shipments Forecast 3 Month Bias

MAD

The MAD function calculates the absolute difference between two measures. Unlike MAPE, which is a percentage value, MAD is a unit value which may be harder to use across an enterprise. It represents a useful value in evaluating areas where forecast and demand differ. Value of MAD is calculated in units, it has no direct indication of accuracy. A product selling a thousand units a day may have a MAD of 100 and be accurate, whereas another product selling an average of ten units a day would be inaccurate with a MAD of 7.

Predefined measures to display MAD are the following:

• Final Bookings Forecast 3 Month MAD

• Final Shipments Forecast 3 Month MAD