Configuring IPM Insights

Configure IPM Insights in Planning to define the insights that planners see on the Insights dashboard.

You configure IPM Insights by selecting the type of analysis to perform, specifying the slice definitions for analysis, configuring the analysis, and then defining the settings for the insights displayed on the Insights dashboard. You can define insights for as many data intersections as you need. Planners see insights only for slices of data to which they have access.

You can also define a prediction using Auto Predict to predict future performance based on historical data and schedule a job to run that prediction definition, automating the prediction process. You can use these prediction results as input for generating insights.

To associate calendars and events with insight definitions and Auto Predict prediction definitions, configure calendars and events first. See Configuring Calendars and Events.

To configure IPM Insights:

1. From the Home page, click IPM and then click Configure.
2. Click Create.
3. In IPM Configurator, on the Types page, enter a name and description.

4. Select Auto Predict to predict future performance based on historical data using time series forecasting techniques. Prediction results can be used as input for generating prediction insights. For more information about Auto Predict, see Setting Up Predictions to Run Automatically with Auto Predict.

5. Select the types of insights to generate, and then click Next. You can select as many as you need.
   • Forecast Bias Insights. Reveals hidden bias in forecasts submitted by planners by analyzing historical data. Measures the variance or bias between two historical scenarios such as Forecasts and Actuals.
   • Prediction Insights. Looks for variance between future scenarios such as forecasts made by planners and computer-generated predictions.

     In Do you have prediction data available?, if you have prediction data available and know where prediction data is, select Yes. If not, select No and Auto Predict is automatically selected. Auto Predict uses the historical data for the intersection you defined to generate prediction data that is used for the insight.

     You can also use a prediction based on Machine Learning instead of auto prediction.

   • Anomaly Insights. Detects outlier values that vary widely from other values.

   You can also choose to just run Auto Predict by selecting Auto Predict.

6. On the Select Calendar page, select a calendar to use, and then click Next. When you select a calendar, the Historical Data section and the Future Data section are populated with details from the selected calendar.

   If you don't select a calendar, you can select these options yourself.

   For Future Data, you can define insights that analyze at the parent level or at the leaf level for Period members. For example, for Future Periods, if you have monthly data, you can perform the analysis at the quarterly level. Note that leaf level and middle level parent members are allowed. Top level parent members, anything directly below Period, are not supported. For Auto Predict, analysis must always be at the leaf level.

7. On the Define Slice page, define the slice of data to analyze for historical data (Actual and Forecast) and future data, and then click Next.

   For considerations on using dynamic calc parent members in the slice definition, see Considerations for Defining the Slice for Insights.

   • In the Historical Data section, select the cube and then define the slice of data for the historical data—both actuals and the forecasts provided by the planners. The slice definitions show all the dimensions for the cube, except for the Year and Period dimensions. All members start out with their root member selected.

   • In the Future Data section, select the cube and then define the slice of data for the future data—both the forecast provided by planners, and the base prediction (most likely scenario).

     If the data is available, click to add a definition for the slice of data for the best case and worst case scenarios.

     If you selected Auto Predict, prediction results are stored in this location.

8. On the Configure page, configure the insights by defining the error tolerance and thresholds for insights, and then click Next. IPM Configurator selects default metrics for analysis.
   • Auto Predict: Select whether to include events in the prediction.

     If you select Show Advanced Options, select additional options to define an auto predict job. For more information, see Configuring Advanced Auto Predict Options.

   • Forecast Variance & Bias Insights: Select an error tolerance percentage, which defines the percentage variance between historical forecasts submitted by planners and actuals that lies within acceptable range. If the percentage variation crosses the error tolerance limit, it is considered for bias calculation.

     For example, if you enter .05, any difference of 5% or above between historical forecasts and historical actuals is considered a significant variance (and considered either under-forecasting or over-forecasting) and triggers an insight.

     You can select or modify the metric and threshold for deviation on the Advanced Options page. See Configuring Additional Insight Settings.

     An insight is generated when either deviation or bias crosses the specified threshold.

   • Prediction Insights: Select a threshold percentage, which defines the acceptable percentage variance between future forecasts submitted by planners and computer-generated predictions.

     An insight is generated when either deviation or risk crosses the specified threshold.

     For example, if you enter 25%, any difference of 25% or above between future forecasts and future predictions would be considered a significant variance (and considered either under-forecasting or over-forecasting). and would trigger an insight.

   • Anomaly Insights: Select whether to include events in the insights calculation.

     Select a threshold, which defines the acceptable threshold for Z-score value (the standard deviation from the mean of the values). Anything that is too far from zero (the threshold is usually a Z-score of 3 or -3) should be considered an outlier.

     An insight is generated when the specified outlier detection metric crosses the specified threshold.

     For example, if you enter 3, anything that is three standard deviations or higher from the mean would be considered an outlier, and would trigger an insight.

   If you want to configure additional options, such as selecting the metrics to use, click Show Advanced Options. See Configuring Additional Insight Settings.

9. On the Settings page, configure the settings that define how to display the insights to planners, and then click Save.
   • Display Dimensions: Select the dimensions that planners will use to navigate and analyze the insights. The dimensions you select will display on the Insights dashboard.
   • Impact Magnitude Thresholds: When insights are displayed on the Insights dashboard, this setting categorizes insights into High, Medium, and Low groups based on the percentage or absolute value impact calculated for each insight. This helps planners focus attention on the insights that have more extreme variations.
     • Anything less than the low value is in the Low category.
     • Anything higher than the high value is in the High category.
     • Anything between the low and high value is in the Medium category.

     For example, if you specify 30% as the Low threshold and 60% as the High threshold, any insight with a percentage impact less than 30% is in the Low category; any insight with a percentage impact greater than 60% is in the High category. Any insight with a percentage impact between 30% and 60% is in the Medium category.

     When entering an absolute value for impact magnitude thresholds, the currency defined in the slice definition applies to the absolute value. If you have multiple currencies defined, you can’t apply an absolute value magnitude threshold; the option is not available.

10. Now you're ready to run the insights you configured. See Running and Scheduling Insights.

Insight definitions are saved as global artifacts (in the Auto Predict folder under Global Artifacts) and are backed up with the maintenance snapshot.

About Using Events with Insights, Auto Predict, and Predictive Planning

Note the following about using events:

• Using events can impact the anomaly insights that are generated. For example, if you create a one-off or repeating event for a specific date where sales spike, and you generate an analysis, anomalies in data for that date are not reported as anomaly insights. Instead, the analysis includes the sales spike for that date. If you generate an analysis without events, an anomaly insight is reported for that date.

• In Auto Predict jobs, note the following behavior for skip and one-off events:
  • For skip events, once the skip event goes into the prediction range, the skip event is not mentioned or listed in the downloaded job report.
  • For one-off events, once the one-off event reaches the last period of the historical data (the period which bumps up against the prediction range), the one-off event is not mentioned or listed in the downloaded job report.
• In prediction on forms using Predictive Planning, note the following behavior for skip and one-off events:
  • For skip events, once the skip event goes into the prediction range, the skip event is no longer applied.
  • For one-off events, once the one-off event reaches the last period of the historical data (the period which bumps up against the prediction range), the one-off event is no longer applied.

Videos

Your Goal

Watch This Video

Learn how to configure insights to automate data analysis. Administrators configure insights to define the insights that planners see on the Insights page. You select the type of analysis to perform, specify the slice definitions for analysis, configure the analysis, and then define the settings for the insights displayed on the Insights page. You can run the insight or schedule it to run on a regular basis.

Oracle Fusion Cloud EPM - Configuring Insights

Tutorials

Tutorials provide instructions with sequenced videos and documentation to help you learn a topic.

Your Goal	Learn How
Learn how to configure IPM Insights by selecting the type of analysis to perform, specifying the slice definitions for analysis, configuring the analysis, and then defining the settings for the insights displayed on the Insights dashboard. You can define insights for as many data intersections as you need. Planners see insights only for slices of data to which they have access.	Configuring Insights