SAP Integrated Business Planning (IBP) for Supply Chain is a dynamic, comprehensive solution that supports various aspects of interactive statistical forecasting and planning:
Immediate demand forecasting (demand sensing): Predicts based on immediate consumer demands.
Mid-term or long-term demand forecasting: Aligns business plans strategically by predicting demands over an extended timeframe.
Quantity forecasting: Anticipates the exact quantity of products needed, eliminating surplus or shortage concerns.
Price and revenue forecasting: Predicts financial aspects, enabling effective financial planning.
The statistical forecasting capabilities of SAP IBP allow for creating forecasts for various scenarios or specific business segments. For example, it can automatically generate quantity forecasts based on shipment history or price and revenue forecasts to predict fluctuations in gasoline prices due to seasonal variations. These tools offer adaptability and predictive accuracy in a changing market environment.
Key Topics
As you progress through this lesson it's important to concentrate on several key topics that are crucial for mastering the content. These areas of focus are not just topics to be learned—they are your stepping stones to a deeper understanding and successful achievement of the course's learning objectives. By emphasizing these crucial areas, you will develop a robust knowledge base that will enhance your proficiency and effectiveness in supply chain management.
Demand Sensing: A method of immediate demand forecasting that predicts consumer demands in real time.
Simple Average: A forecasting algorithm that calculates the mean of all historical data within a specified time horizon.
Simple Moving Average: A forecasting algorithm that calculates the mean of subsets of historical data over a specific number of periods.
Single Exponential Smoothing: A forecasting algorithm that assigns exponentially decreasing weights to older data points, useful for data without trend or seasonality.
Double Exponential Smoothing: A forecasting algorithm that manages both trend and key figure values, useful for data with a trend.
Triple Exponential Smoothing: A forecasting algorithm that manages trend, seasonality, and key figure values, useful for data with both trend and seasonality.
Pre-Processing Steps: Initial steps in a forecast model that address issues in historical data before generating forecasts.
Post-Processing Steps: The final steps in a forecast model, focusing on actions after forecast generation, including error measurement.
Forecast Model Algorithms in SAP IBP for Sales and Operations
The core of statistical forecasting is the predictive capabilities of diverse forecast models. SAP IBP continually enhances its functionalities by introducing new models for increased utility and precision. The 'Manage Forecast Model' application is where these models are defined. Combining multiple robust algorithms elevates business prediction precision.
However, it's crucial to choose a specific perspective for your algorithms. Short-term and mid-term or long-term forecasting algorithms cannot coexist within a single model. This specificity ensures accurate and beneficial results.
Understanding the statistical forecasting algorithms included in the SOP license can significantly enhance demand planning effectiveness, allowing for predictive strategies with proficiency and confidence.
Note
You can only use SAP IBP application algorithms that your company has licensed.Forecasting Algorithms
Forecasting algorithms are sophisticated mathematical methods that help project future product demand based on historical time series values. Here’s an overview of the algorithms used within the Sales and Operations Planning (S&OP) solution:
Simple average:
Computes the mean of all historical values within a specific time horizon, using it as a future prediction. The forecast is a uniform value based on weekly or monthly history.
Simple moving average:
Calculates the mean of time series values in successive subsets of time periods. The number of periods in these subsets is the key parameter.
Single exponential smoothing:
Models a time series without trend or seasonality, assigning less importance to older data through exponentially diminishing weights (alpha coefficient between 0 and 1).
Double exponential smoothing:
Models time series with trend or seasonality, managing trend and key figure values using alpha and beta coefficients (both between 0 and 1).
Triple exponential smoothing:
Manages time series with trend and seasonality, minimizing their influence on projections using alpha, beta, and gamma coefficients (all between 0 and 1).
Copy past periods
Useful for recurring patterns, this algorithm suits scenarios where monthly sales data mirrors previous years' trends.
Using Multiple Algorithms.
When incorporating multiple forecasting algorithms into a model, the system's engagement with each algorithm can be controlled:
Choose best forecast:
The system uses a specified error measure to identify and apply the most accurate forecast algorithm.
Calculate weighted average forecast:
Weights are assigned to forecasting algorithms, and an average forecast is computed by multiplying each result by its corresponding weight.
Note
For more information, see http://help.sap.com/.Manage Forecast Models
The Manage Forecast Models application allows for creating and modifying models that underpin forecasting processes within SAP IBP. Users can outline pre-processing, forecasting, and post-processing steps within distinct tabs for straightforward navigation.
General settings
Establish parameters applicable to the entire model.
Pre-Processing Steps:
Select algorithms to mitigate issues in the time series data before generating forecasts.
Post-Processing Steps:
Select settings for actions performed by the system after forecasting, including error measure computations.
Video Summary
When defining a statistical forecasting model, the user has flexibility in three key areas: pre-processing, which involves data preparation; forecast steps, which include running the forecast; and post-processing, which involves tasks such as saving error measures. These sections are organized as tabs in the statistical models configuration screen, allowing for a clear and structured setup process.
You can choose settings that are valid for the entire model regardless of the algorithms selected by selecting General.
By selecting Pre-Processing Steps, you can select algorithms that are used to solve possible issues in the time series data before the forecast is calculated. The following types of issues can occur:
Some key figure values may be missing from the input data.
The input data may contain unusual values (outliers).
By selecting Forecasting Steps, you can choose the settings that define how the system calculates the forecasts for the selected key figures. Forecasts are calculated both for the past and future with the help of the selected algorithms.
By selecting Post-Processing Steps, you can choose the settings for the steps that are performed by the system after forecasting. You can select methods for calculating error measures.
