Build Anomaly Detection (BETA)

Edit online

The anomaly detection feature is designed to help you identify unusual changes in calculation effort, which can help you optimize your reports and transforms. This feature is based on the data collected by Calc Explorer, which provides detailed information about the calculation effort for each build. This feature is currently behind a BETA flag.

The anomaly detection feature uses the data from Calc Explorer to identify unusual changes in calculation effort. It compares the calculation effort for each build to the previous builds and identifies any significant changes. If a significant change is detected, the feature will alert you and provide detailed information about the change.

Benefits of Anomaly Detection

The anomaly detection feature can help you:

  • Identify and optimize inefficient reports and transforms
  • Reduce calculation effort and improve performance
  • Improve the overall quality and reliability of your reports and transforms

How to Use Anomaly Detection

To enable the anomaly detection feature, do the following:
  1. Navigate to Project > Enable Features and select Show Anomalies Panel checkbox.
  2. Navigate to TBM Studio > Project > Anomalies tab.

    img

  3. Expand the anomaly type and investigate any anomalies that are detected.

    img

    Table 1.
    Field Description
    Started Shows the timestamp when the calc started
    Calc Explorer Links to the calc which had the anomaly
    Change Set Links to the list of transforms and entities changed with the build
    Changes Links to the change messages associated with the build
  4. Take appropriate action to optimize the reports and transforms as needed.

Types of Anomalies

There are two types of anomalies that the feature can detect:

  • Calculation Anomaly: This type of anomaly is an increase in calculation effort over what what was expected for a calc.
  • Entity-Based Anomaly: This type of anomaly occurs when there is an unusual pattern in the data shape (e.g., an unusual number of rows).
  • Combined: This type of anomaly occurs when there is a combination of calculation and entity-based anomalies.

Calculation Anomalies

A calculation anomaly occurs when the calculation time for a build is significantly higher than usual. This can be caused by various factors, such as:
  • Uploading a large amount of data, modifying formulas, or changing report configurations.
  • Changes that increase the calculation effort, such as adding new reports or transforms.

Investigating Calculation Anomalies

To investigate a calculation anomaly, you can:
  • View the change set: See the specific changes made in the build, such as uploaded files or modified formulas.
  • Explore Calc Explorer: Analyze the calculation effort for the build and identify areas where the calculation time is higher than usual.
The system will eventually provide explainability features to help you understand the cause of the calculation anomaly. This will include information such as:
  • Specific files or changes: Which files or changes caused the calculation time to increase.
  • Assignment ratios table: How changes to the assignment ratios table affected the calculation time.

Troubleshooting

If you encounter any issues with the anomaly detection feature, do the following:

  • Check the Calc Explorer data to ensure that it is accurate and up-to-date.
  • Investigate any anomalies that are detected and take action to optimize the reports and transforms as needed.

FAQs

  • Q: What is the purpose of the anomaly detection feature? A: The anomaly detection feature is designed to help you identify unusual changes in calculation effort, which can help you optimize your reports and transforms.
  • Q: How does the anomaly detection feature work? A: The anomaly detection feature uses the data from Calc Explorer to identify unusual changes in calculation effort. It compares the calculation effort for each build to the previous builds and identifies any significant changes.
  • Q: What types of anomalies can the feature detect? A: The feature detect changes in calculation effort due to material changes, increased calculation effort, or data shape anomalies such as outliers, skewness, or distribution shifts.