Setting up Alerts for Forecast Accuracy

Custom alerts are used to find important or critical supply chain issues such as inventory shortages, an imbalance of supply and demand, any unexpected changes in the supply chain or low forecast accuracy.

You can specify the threshold values to be used to determine issues. For example, you can specify the threshold values for minimum stock levels in a particular location.

Based on these threshold values, the system analyzes the data on the fly and finds where the threshold values are reached. This enables you and your colleagues to react quickly, before a supply chain situation becomes a problem.

For analysis, it is not only analytics and dashboards that can be used, but also custom alerts created by users or IT administrators. It is better to create custom alerts for forecast accuracy after ABC/XYZ segmentation because that provides an opportunity to check alerts per segment.

Custom alerts are generated by the system and visible to assigned users regarding important or critical supply chain situations. Information contained in the custom alert allows users to analyze and measure the impact of unexpected situations on their business. Users can fine-tune the criteria for alerts to be generated to mitigate issues in the supply chain. Custom alerts are integrated with cases in the SAP Supply Chain Control Tower, which facilitates the tracking and resolution of supply chain problems.

It is necessary to define the planning area, calculation levels, and time horizon you would like to use as your data set to trigger alerts.

You can do the following:

• Add rules and conditions under which alerts are raised.

• Specify the minimum number of consecutive periods that a condition must exist before an alert is generated.

• Include the metrics that you want to be displayed to help you analyze the alert and take action.

• Choose for which version of the data you would like to trigger alerts.

• Make a setting for alerts that may occur across an established time horizon to be aggregated into a single chart display in the monitor.

• Select various options for displaying the alert data (using different chart types).

• Subscribe to custom alert definitions and add filters if needed to restrict or further customize the alerts that are triggered.

  Note

  Only alert definitions that have been subscribed to, generate alerts in the Monitor Custom Alerts app.

• Assign an Excel template so that users can solve alerts in Microsoft Excel.

• Share definitions and subscriptions with other users or user groups.

• Opt out of definitions and subscriptions that are shared by others if you are no longer involved in resolving them.

• Navigate directly from the definition to the monitor.

It is necessary to define rules for any alerts to be generated. Rules are based on a combination of key figures upon which the application calculates on the fly the alert situations.

You can set the condition that all rule groups (you must have at least one rule group) must satisfy (default setting) or, if you have more than one rule group, any or all groups must satisfy for an alert to be triggered.

Within the rule group, you have individual rules (you must have at least one rule). You can set the condition that all the rules must be satisfied (default setting) or, if you have more than one rule, any or all of the rules must be satisfied for an alert to be triggered.

You can define rules in either of the following ways:

• Absolute: One key figure that is compared to a threshold value.

• Percentage (%): Two key figures, where both are compared using a percentage value.

You can use the following operators:

• = (is equal to)

• ＜ (is less than)

• ＞ (is greater than)

• ＜=(is less than or equal to)

• ＞=(is greater than or equal to)

• ＜＞ (is not equal to)

• is null

Null is used only to compare one key figure. For example, you create a rule as follows: Confirmed quantity is null. The system returns records for which it finds for the confirmed quantity no values in the database. This is to differentiate between null and zero values.

You can apply two clustering-based algorithms, k-means and Density Based Spatial Clustering of Applications with Noise (DBSCAN). The number of alerts can be either reduced or increased to have meaningful insights. Also, if you change data, the alert definition is automatically adjusted.

Using the standard method, it may be cumbersome to create your alert definition rules to ensure that the correct outliers are identified. Often, this leads to creating several alert definitions to determine the one outlier. With machine learning, this problem is resolved as the algorithm carries out optimal clustering to determine the outlier.

Custom alerts are based on static rules. This works fine if the threshold for the exception condition is known and the data is generally consistent. If the data is variable and changes to the pattern occur, the static rules may lead to too many or too few alerts, which either keep planners busy or leaves them clueless. With machine learning, planners can set up alerts without needing to know the exact thresholds.

For example, if you define alerts for your product’s sales quantity based on an ABC indicator, the rules would be different for A product (fast seller), which could be set at 5% decline in sales, than and for the C product (slow seller), which could be set for 20% decline in sales. This would result in several different alerts to provide the same result, which is to indicate that there has been a drop in your sales quantity.

With machine learning, the grouping of your ABC products can be done automatically. Similarly, it allows you to find the outliers in the groupings.

The k-means and DBSCAN clustering methods are used to find and group points on a chart that are close together. They also help in identifying the outliers (isolated points in low-density regions outside the groupings).

This cluster analysis method is popularly used in machine learning. It is simpler to understand and execute, but less accurate for finding the outliers.

The k-means algorithm requires a minimum number of clusters to be provided and the clusters are expected to be of similar size. However, the number of clusters in SAP Integrated Business Planning is determined automatically.

The DBSCAN algorithm checks attribute groupings and performs clustering on these attributes. In addition to performing outlier determination on one key figure, it uses multiple attributes during the calculation process that makes it more accurate than the k-means algorithm.

You can use the Define and Subscribe to Custom Alerts app to create subscriptions to custom alert definitions that allow you to view any alerts triggered in the Monitor Custom Alerts app. The custom alerts in the alert monitor are based on all the custom alert subscriptions that you have created or the ones that have been shared with you.

You can perform the following tasks:

• Define which metrics are displayed in the Monitor Custom Alerts app for the alert.

• Specify the number of periods before and after the alert occurrence that you want to have displayed in the Monitor Custom Alerts app.

• View the number of alerts that this subscription generates.

• Add filters to refine the alert data or remove filters as needed.

• Delete or deactivate your own subscriptions so that you are no longer subscribed to a custom alert definition, which can be helpful if your job responsibilities have changed or there is an issue with the data.

• Opt out of the subscription.

• Share the subscription with users and user groups.

You can perform the following tasks in Monitor Custom Alerts app:

• Analyze the charts and metrics to identify and resolve potential issues.

• Link a new case or an existing case to an alert to follow up on the issues, delegate them, and resolve them.

• Evaluate and assess the risks you may encounter if you do not take immediate action.

• Filter custom alerts by case.