Amazon QuickSight
User Guide

Using ML-Powered Anomaly Detection

Use the following procedure to start detecting anomalies in your analysis.

If you already have anomaly detection in your analysis, skip ahead to the following section to learn how to explore the anomalies detected by Amazon QuickSight.

  1. Begin with an existing analysis. Choose one of the following:

    • On the left, choose Insights. Then choose Add anomaly to sheet.

      This creates a widget entitled Anomaly detection. Follow the screen prompts to add fields.

    • On the top menu, choose Add, then Add insight. From the list, choose Anomaly detection and Select.

      This creates a widget entitled Insight. Follow the screen prompts to add fields.


    In the field wells, Categories represent the dimensional values that Amazon QuickSight splits the metric by. For example, let's say you are analyzing anomalies on revenue across all product categories and product SKUs. There are 10 product categories, each with 10 product SKUs. Amazon QuickSight splits the metric by the 100 unique combinations and runs anomaly detection on each for the split metric.

    You can add up to five category fields to an anomaly detection job. Table calculations don't work with anomaly detection.

  2. Choose the widget that you added to your analysis. Then choose the Get started button on the widget.

    A scrollable screen appears with configuration settings for anomaly detection.

  3. On the scrollable screen, configure one or more of the following:

    • Name – provide a descriptive name for your anomaly detection.

    • Fields for analysis – view the contents of the field wells. To edit these, choose Cancel to exit this screen, and then add your new fields.

    • Analyze all combinations of these categories – by default, if you have selected three categories, Amazon QuickSight runs anomaly detection on the following combinations hierarchically: A, AB, ABC. If you choose this option, Amazon QuickSight analyzes all combinations: A, AB, ABC, BC, AC. If your data isn't hierarchical, make sure to enable this option.

    • Number of anomalies to show – set how many anomalies you want to display on the narrative widget. You can still explore all the anomalies, no matter how few you choose to show in the analysis.

    • Sorting method – choose the method you want to apply to sorting anomalies. You can use any of the following options.

      • Weighted anomaly score – the anomaly score multiplied by the log of the absolute value of the difference between the actual value and the expected value. This is always a positive number.

      • Anomaly score – the actual anomaly score assigned to this data point.

      • Weighted difference from expected value – (default) the anomaly score multiplied by the difference between the actual value and the expected value.

      • Difference from expected value – the actual difference between the actual value and the expected value (actual−expected).

      • Actual value – the actual value with no formula applied.

      Schedule – set the time interval at which anomaly detection recalculates. The schedule runs only for published dashboards. In the analysis, you can run it manually in the analysis as needed.

      • Occurrence – set how often you want the recalculation to run: every hour, every day, every week, or every month.

      • Start schedule on – set the date and time to start running this schedule.

      • Timezone – set the time zone that the schedule runs in.

    • Contribution analysis – (optional) analyze the top contributors when an anomaly is detected. For example, Amazon QuickSight can show you the top customers that contributed to a spike in sales in the USA for home improvement products. If you have additional dimensions in your data—dimensions you aren't already using in the anomaly detection—you can add them here for contribution analysis.

    • Dimensions – view a list of dimensions to choose contributors from. You can choose up to four dimensions.

  4. Choose OK to confirm you choices. Choose Cancel to exit without saving. Choose Delete computation to remove it.

    To reopen the configuration screen, choose the v-shaped on-visual menu, then choose Configure anomaly.

  5. Run the anomaly and see your anomaly narrative by choosing Run now.

    The amount of time it takes to complete anomaly detection varies depending on how many unique data points you are analyzing. The process can take a few minutes for a minimum number of points, or it can take many hours. While it's running in the background, you can do other work in your analysis. However, you shouldn't change the anomaly detection configuration or narrative, or explore the anomalies during run time.

  6. (Optional) Remove the anomaly detection by choosing the v-shaped on-visual menu, then choosing Configure anomaly. Then choose Delete computation at the bottom of the screen. To confirm deletion, choose Delete. Otherwise, choose Cancel.