Forecast and Forecast accuracy

Forecasting plays a crucial role in supply chain management, helping organizations predict future demand, plan production schedules, optimize inventory levels, and allocate resources efficiently. A forecast is an estimate or prediction of future demand for products or services based on historical data, market trends, and other relevant factors.

Forecasting methods in majority are of following types:

i - Qualitative - Based on human judgement/gut feeling/experience.

ii - Time Series - based on historical demand/data.

iii - Causal - interlinked with certain environmental factors like interest rate, economy etc.

iv - Simulation - combined form of time series and causal methods.

Time series is the most widely method used.

Forecast Error, also known as Forecast Accuracy, is a quantitative measure that indicates the difference between the forecasted values and the actual outcomes. It represents the degree of deviation between the predicted forecast and the realized demand or sales.

Forecast Error is essential for evaluating the effectiveness of forecasting models, identifying areas for improvement, and enhancing the accuracy of future forecasts.

Common metrics used to measure Forecast Error include:

i.                  MAPE (Mean Absolute Percentage Error)

ii.                 MAD (Mean Absolute Deviation)

iii.                MSD (Mean Squared Deviation)

iv.                RMSE (Root Mean Square Deviation)

v.                 ME (Mean Error)

vi.                MPE (Mean Percentage Error)

Method 1: Mean Absolute Percentage Error (MAPE) in %

Use: MAPE is a type of percentage error. It measures the error in %.

MAPE =1/n∑[(∣A-F∣)/(∣A∣)]* 100

n = number of period

A = Actual Order Qty in the month M

F = Forecast Qty in the month M-2 or M-3 for month M

Example:

Period	Material	OOH(A)	Forecast (F)	∣A-F∣	∣A-F∣/∣A∣
Jan	AXX	50000	42000	8000	.16
Feb	AXX	6900	5600	1300	.19
Mar	AXX	300	450	150	.50
Apr	AXX	700	600	100	.14
May	AXX	1800	2300	500	.27
n = 5					∑ = 1.26

MAPE=(1.26/5)*100 = 126/5 = 25%

Accuracy = 75%

Note: I will propose this model (MAPE) and calculation, as other models are giving deviation in absolute term and not in %.

Method 2: Mean Absolute Deviation (MAD)

Use: MAD measures the size of the error in units. It gives absolute value of the error irrespective of sign.

MAD =1/n∑ ∣Actual - Forecast∣

n = number of period

A = Actual Order Qty in the month M

F = Forecast Qty in the month M-2 or M-3 for month M

Method 3: Mean Squared Deviation (MSD)

Use: MSD measures the size of the error in units. It gives absolute value of the error irrespective of sign. It is more sensitive than MAD.

MAD =1/n∑ (∣Actual - Forecast∣) ²

n = number of period

A = Actual Order Qty in the month M

F = Forecast Qty in the month M-2 or M-3 for month M

Method 4: Mean Error (ME)

Use: ME is a metric used to measure the average difference between forecasted values and actual observed values.

It provides insight into the overall bias or tendency of a forecasting model to overestimate or underestimate the actual values.

ME=1/n∑ (A – F)

While Mean Error provides a simple measure of bias in forecasting models, it is often used in conjunction with other metrics to assess overall forecasting performance comprehensively.

Method 5: RMSE (Root Mean Squared Error)

Use: RMSE is a commonly used metric to evaluate the accuracy of a forecasting model.

It provides a comprehensive measure of the forecast error, considering both the magnitude and direction of deviations.

RMSE = √(1/n)∑ (A - F)

There are other methods too like Relative percentage difference, Z score, modified Z score which are applicable case to case.