Anomaly details

Overview

• The Anomaly Details modal provides specific details about the Anomaly, including source record and checks that failed assertions, along with ability to take Notes and provide feedback to ML methods through updating Status of the anomaly.

• Users have the ability to provide feedback to the ML methods through Supervised Learning. Specifically, the user can Acknowledge, Resolve or Invalidate an anomaly. Each of these actions will change the tolerances of the data quality checks behind the anomaly.

Datastore details

1. The # ID of the anomaly: 32102
2. The Datastore name: SAP Financial Exports
3. The Location where this data is stored: s3a://qualytics-demo-data/orders/demo/customer-20220223.csv
4. The Record Type of the anomaly: Shape
5. The Tags are labels that serve the purpose of grouping anomalies and driving downstream workflows: Migration Low
6. The Status of the selected Anomaly: Users can edit the status to Acknowledged, Resolved or Invalid.
7. The Detected at of the anomaly creation time: 1 year ago

Failed Fields details:

Field

All the fields of that specific anomaly found: company_name

Rule

Rule type that failed the assertion(s)

You can check all the Rule types here.

Violation

Details of how the anomaly failed the assertion(s) of rule(s)

Coverage

The expected tolerance of the rule

Type

Infered or Authored

Coverage

The expected tolerance of the rule

Tag

The rule type Tag

Source Records

Show in a tabular view all the records and fields and highlight records with anomaly data based on the rule type

Users can filter by Field in this view

Infered check details

• Clicking into a Rule will highlight its details and enables users to edit the rule as well:

• An anomaly can be archived pressing the Check Actions menu and Archive.

• If you expand the section Advanced Options you can add a Filter clause and also change the Coverage percentage for that anomaly.

Note

The Filter clause is a WHERE statement against your table. For example:
price != 33 or price > discount + 20

Info

You can create a computed table and use multiple fields from different tables in a filter clause

Suggested remediation

Qualytics can also provide you a suggested value

Download Source Records

Download the source records for further analysis or external use.

---

Last update: April 7, 2024