In order to carry out the forecast for a material, you must first of all define the forecast parameters for this material in the material master record.

You can only carry out the forecast if historical consumption values for this material are available. In general, the system updates the consumption values in the material master record during withdrawal postings. Before the forecast is carried out, the consumption values can be checked and corrected if necessary. In addition, you can manually enter historical consumption values as basic values for the first forecast.

If no historical data exists for a material which the system can access during the forecast, you can carry out the forecast with reference to another material.

The forecast results are only used for MRP if the forecast indicator is set in Customizing.

When a series of consumption values is analyzed, certain patterns can usually be detected. From these patterns it is then possible to differentiate between various forecast models:

Constant model: A constant consumption flow applies if consumption values vary very little from a stable mean value.

Trend model: With a trend model, consumption values fall or rise constantly over a long period of time with only occasional deviations.

Seasonal model: If periodically recurring peak or low values which differ significantly from a stable mean value are observed, it is a case of a seasonal consumption flow.

Seasonal trend model: A seasonal trend consumption model is characterized by a continual increase or decrease of the mean value.

If none of the above patterns can be detected in a series of past consumption values, then we have an irregular consumption flow.

Before the first forecast can be carried out, you must define which forecast model the system is to use to calculate the forecast values.There are three basic possibilities:

Manual model selection,

Automatic model selection,

Manual model selection with additional system check.

If you want to select a model manually, then you must first of all analyze past consumption data.

If you want to have a model automatically selected, you can instruct the system to analyze the historical data with respect to trend model, seasonal model, or seasonal trend model.

If you want to use manual model selection with additional system check, you specify a model manually and set the system so that it additionally checks historical values for a seasonal pattern or trend.

During the first material forecast, the necessary model parameters (basic value, trend value, seasonal indices) are determined for the respective forecast model. Model initialization takes place during the forecast. It must also be carried out in cases of structural interruption, that is, if the existing forecast becomes invalid.

Parameter optimization is used to optimize the smoothing factors. The system calculates several parameter combinations and then selects the combination with the lowest mean absolute deviation.

Ex-post forecast is a forecast for past periods. A forecast for the latest past consumption values is carried out on the basis of older past consumption values. They are compared to the actual consumption values in order to adjust the model parameters to the actual consumption flow in an optimum way.

If more past values are available than the system requires or is to use for model initialization, then the system will carry out an ex-post forecast. In order to do this, the system divides the time series of past values into two groups. The first group with the older values is used for initialization. The forecast is repeated for the second section, that is, the forecast is carried out parallel to the time series of consumption. The forecast values for the past are then compared to actual consumption values. Thus, the parameters are adapted to the most recent developments.

The ex-post forecast is used for:

Models based on exponential smoothing

Parameter optimization

Skipping forecast periods

Evaluation of the forecast accuracy

There are several possibilities available to you for carrying out the material forecast:

Individual forecast (interactive/not interactive)

Total forecast (online/in background mode)

The interactive forecast is triggered in the material master record from the forecast screen. The forecast settings can be changed interactively as often as necessary before the forecast results are saved. An interactive graphic is available for checking the forecast where the forecast values can be changed.

The total forecast is carried out at plant level and for each period indicator. In order not to overload the system, you should carry out the total forecast in background mode. As soon as the forecast run is finished, the system displays a list of all materials for which the forecast was carried out, provided you have set the indicator Log record. If not, no list will be generated.

After a forecast run, the results must be checked over to make sure that the forecast was carried out correctly. There are two ways of doing this:

Checking the forecast result using a listYou can print a list of the materials which were included in the forecast via a print program and then you can use this list to check the results.

Checking the forecast result in online modeErrors or exceptional situations which appear during the total forecast are recorded in the exception messages and are allocated to a certain error class. If an error occurs in a material forecast, then the system marks the material for which the error occurs for reprocessing. The marked materials can be checked and reprocessed in online mode. If you want to limit the selection, you can specify an error class.

Possible exceptional situations include the following:

No historical data exists

Safety stock not calculated as service level = 0,

Pattern of consumption series has changed.

The allocation of an exceptional situation to an error class is carried out in Customizing.

The historical data updated by the system cannot be changed manually but can only be influenced by correction values. This is for example necessary for outliers.

The forecast values calculated by the system also cannot be changed manually but can only be influenced by correction values.

The system automatically sets a firming indicator for corrected forecast values. This means that the correction value cannot be changed during subsequent forecast runs.

Forecast values can also be firmed manually. In this case, the values also cannot be changed during subsequent forecast runs.

The number of historical data defines the scope of historical data the system has to consider during the forecast. The default value is the maximum possible number (60). If you limit the number of historical data, only the latest values are taken into account during the forecast.

When calculating the forecast, the system also calculates the degree of error (error total, MAD, tracking signal).

The forecast error is the difference between actual consumption values and the forecast values from the same period whereas the error total is the sum of all the forecast errors in a consumption series. The error total is used to check the validity of the forecast model in operation. If a model is still valid, you can assume that the error total is distributed normally and has an average of zero.

If the consumption pattern has changed, however, the error total will no longer be equal to zero. You must standardize the calculated forecast error in order to set boundaries. Therefore, in addition to the error total, the system also calculates the mean absolute deviation (MAD) as a second value. The system then adds the errors (irrespective of the plus or minus sign) and divides them by the number of consumption values.

Using the quotient from the error total and the MAD, the system can now define a tracking signal during every forecast to identify changes in the pattern of a consumption series in time. The tracking signal is compared to the tracking limit specified in the material master. When the tracking signal is greater than the tracking limit, you will receive a message stating that the forecast model should be checked.

The system automatically sets the tracking limit (the default value is 4.00). However, it can be changed by the MRP controller in the material master record.

SAP AGTAMM10 4.0B36-13