To receive anomaly notifications for data that is relevant and applicable for business users, you can apply filters on the data. The filters narrow down the data elements for which data observability sends notifications to users in Data Governance and Catalog.
1Open the catalog source for which you want to configure data observability.
2In the Configuration tab of the wizard, go to the Data Profiling and Quality tab.
The options to configure data profling, quality and observability appear.
3Enable and configure data profiling for the catalog source.
4In the Data Observability section, enable data observability.
The configuration parameters for data observability appear.
5Configure the parameters for data observability.
The following table lists the parameters you can configure:
Option
Description
Minimum Number of Data Points
Specify the minimum number of profiling runs that are required for data observability to start detecting anomalies. For example, if you enter 4 here, anomalies are detected for 4 and subsequent profiling runs. The default value is 3.
Enter a number between 3 and 5.
Metric Filters
Select an option to indicate whether you want to further filter the profiled data elements.
- No filters. Do not filter the profiled data elements. Data observability detects anomalies on the data that you have configured for metadata extraction and profiling.
- Filter conditions: Select one or more conditions to filter the profiled data elements. For data observability to detect anomalies, create a further subset of data after metadata extraction and profiling.
Inclusion or exclusion criteria
Select the filter condition to apply on the profiled data.
- Include Metric. Specify an inclusion criteria. Data observability detects anomalies on the profiled data that meets the filter criteria.
- Exclude Metric. Specify an exclusion criteria. Data observability excludes profiled data that meets the filter criteria.
You can further narrow down the results by clicking Add to add further filter conditions.
Metrics
Select the metric for which data observability notifies users of anomalies.
Sensitivity
Select the sensitivity of the anomaly.
- Normal. Data observability notifies users of anomalies about normal changes to data.
- Sensitive. Data observability notifies users of anomalies about sensitive changes to data.
- Severe. Data observability notifies users of anomalies about severe changes to data.
Detection rules
Select one or more rules to apply on the profiled data to detect anomalies.
- Static Data. Detect the following anomalies:
- Percentage variation
- Count variation
- 100% or 0% Change Detection. Detect the following types of percentage-based anomalies:
- Drop from maximum
- Surge from minimum
- Standard Deviation. Detect the following anomalies:
- Drop in transition
- Surge in transition
- Deviation
- Breaking Trends. Detect the following types of count-based anomalies: