August 16, 2016 at 11:08The art of forecasting in SAP F&R inventory management
SAP F&R (Forecasting & Replenishment) is a system of order planning and demand forecasting for the formation of order projects at the store-supplier level. The system is part of the SAP SCM (Supply Chain Management) solution and is implemented in two variations:
This post discusses the possibilities of calculating the average forecast in the SAP F&R system.
The average forecast in SAP F&R terminology is such a value of the volume of goods at a location that with a probability of 50% will satisfy the demand of customers in the store, or, in other words, will ensure the level of customer service = 50%.
In order to avoid lost sales, the target level of customer service, as a rule, is planned at least 95%. This means that in 95 cases out of 100, the customer will buy what he planned in the store. Providing a high level of service in SAP F&R is carried out by means of an insurance premium to the average forecast, which depends not only on the target level of service, but also on the variability of the past sales values of the goods. Thus, the maximum sales forecast is generated in the system, the volume of which will be enough to minimize the stock in the warehouse or store (and, therefore, the withdrawal of capital frozen in stocks) and comply with the target level of customer service.
To build an average forecast, the SAP F&R system uses sales forecasting models that take into account not only static consumption data, but also the influence of external factors, such as calendar events or promotions. The effect of such factors can be set either manually or automatically detected by the system in the past. Thus, in the formation of the forecast model of the time series both in the past and in the future, SAP F&R uses data on a possible change in the predicted value and imposes the effect of an external factor on the smoothed series.
As you can see in the figure, when forming the forecast model in the past, SAP F&R clearly revealed seasonal fluctuations in sales and peak surges in the New Year period. However, not every historical data changes naturally, and very often when forecasting, you may encounter the situation below:
How, then, to consider the average forecast so that its reliability is sufficiently high?
Such time series are a random Poisson distribution, so the probability that sales will be m pieces is calculated by the formula:
Despite the fact that the Poisson series is very easy to see - they are quite difficult to determine. In the above example, the value of m is the best predicted value (forecast A). Zero predicted values (or unit for the last case) - are the medians of this time series, i.e. 50% probability of satisfying customer demand (forecast B).
To determine the most relevant forecast value, we will check the forecast error when using the Poisson distribution and the median for calculating average sales using the wMAPE formula - the weighted average absolute error in percent:
Thus we get the forecast error:
In each case, the forecast with a wMAPE error is lower than expected. Therefore, the use of wMAPE in this case almost always leads to an incorrect result.
Then we resort to calculating the errors using the formulas MSE (root mean square error) and sRMSE (root of the root mean square error). To calculate the root mean square error, we use the formula:
The results are as follows:
In this case, using the probability of obtaining a value from the Poisson distribution for forecasting is more preferable. Now we calculate the error using the sRMSE formula using the formula:
The results are as follows:
Thus, it is obvious that the systematic use of error calculation using the wMAPE method leads to a decrease in the accuracy of the forecast, and when a Poisson distribution is detected, it is impractical to estimate the error by such methods as wMAPE, MAE (mean absolute error), MAD (mean absolute deviation), MASE ( average absolute scaled error).
A similar effect is often manifested when forecasting sales of goods with a low turnover rate (less than 0.2 units per day). With this turnover, the best methods for estimating sales forecast errors are MSE and sRMSE. At the same time, it is necessary to remember the constant monitoring of the level of inventories and the assessment of the shortage of goods, because the accuracy of forecasting is just a means.
- SAP F&R SCM - implementation with seamless integration with SAP systems;
- SAP F&R OI is a system for integration with non-SAP systems.
This post discusses the possibilities of calculating the average forecast in the SAP F&R system.
The average forecast in SAP F&R terminology is such a value of the volume of goods at a location that with a probability of 50% will satisfy the demand of customers in the store, or, in other words, will ensure the level of customer service = 50%.
In order to avoid lost sales, the target level of customer service, as a rule, is planned at least 95%. This means that in 95 cases out of 100, the customer will buy what he planned in the store. Providing a high level of service in SAP F&R is carried out by means of an insurance premium to the average forecast, which depends not only on the target level of service, but also on the variability of the past sales values of the goods. Thus, the maximum sales forecast is generated in the system, the volume of which will be enough to minimize the stock in the warehouse or store (and, therefore, the withdrawal of capital frozen in stocks) and comply with the target level of customer service.
To build an average forecast, the SAP F&R system uses sales forecasting models that take into account not only static consumption data, but also the influence of external factors, such as calendar events or promotions. The effect of such factors can be set either manually or automatically detected by the system in the past. Thus, in the formation of the forecast model of the time series both in the past and in the future, SAP F&R uses data on a possible change in the predicted value and imposes the effect of an external factor on the smoothed series.
As you can see in the figure, when forming the forecast model in the past, SAP F&R clearly revealed seasonal fluctuations in sales and peak surges in the New Year period. However, not every historical data changes naturally, and very often when forecasting, you may encounter the situation below:
How, then, to consider the average forecast so that its reliability is sufficiently high?
Such time series are a random Poisson distribution, so the probability that sales will be m pieces is calculated by the formula:
Despite the fact that the Poisson series is very easy to see - they are quite difficult to determine. In the above example, the value of m is the best predicted value (forecast A). Zero predicted values (or unit for the last case) - are the medians of this time series, i.e. 50% probability of satisfying customer demand (forecast B).
To determine the most relevant forecast value, we will check the forecast error when using the Poisson distribution and the median for calculating average sales using the wMAPE formula - the weighted average absolute error in percent:
Thus we get the forecast error:
In each case, the forecast with a wMAPE error is lower than expected. Therefore, the use of wMAPE in this case almost always leads to an incorrect result.
Then we resort to calculating the errors using the formulas MSE (root mean square error) and sRMSE (root of the root mean square error). To calculate the root mean square error, we use the formula:
The results are as follows:
In this case, using the probability of obtaining a value from the Poisson distribution for forecasting is more preferable. Now we calculate the error using the sRMSE formula using the formula:
The results are as follows:
Thus, it is obvious that the systematic use of error calculation using the wMAPE method leads to a decrease in the accuracy of the forecast, and when a Poisson distribution is detected, it is impractical to estimate the error by such methods as wMAPE, MAE (mean absolute error), MAD (mean absolute deviation), MASE ( average absolute scaled error).
Conclusion
A similar effect is often manifested when forecasting sales of goods with a low turnover rate (less than 0.2 units per day). With this turnover, the best methods for estimating sales forecast errors are MSE and sRMSE. At the same time, it is necessary to remember the constant monitoring of the level of inventories and the assessment of the shortage of goods, because the accuracy of forecasting is just a means.