Predictive AI Testing

The Fairly platform performs predictive AI testing using a combination of statistical and industry best-in-class techniques and methodologies.

Challenges for Predictive AI testing often include lack of access to training data and proprietary model algorithms. The use of BISG and relying on input and exhaust data for analysis helps address these issues.

How does Asenion Test Agents perform these tests?

• Provide Dataset File: Enter the absolute path to your dataset file on your local env.

• Provide Dataset Configuration File: Enter the absolute path to your dataset configuration file on your local env. This is a YAML file that maps the columns in your dataset to the column structures understood by the test agent.

• Provide Fairly AI Platform URL: Enter the URL to the Fairly AI Platform you are using.

• Provide Security Token: Enter the security token for the Fairly AI Platform. (You can generate your security token by logging into your Fairly AI Platform account -> Settings -> Generate Access Token)

• Provide Project ID: Enter the project ID of the project you want to upload the results to.

• Processing Files: After you click the ‘Process File’ button, the application will display a message saying “Processing files, this may take a minute…” and then start processing the files.

• Running Tests: The application runs a series of tests on the processed files. The output of these tests is captured and will be displayed in the application.

• Displaying Results: After the tests are completed, the application will display a message saying “Files processed successfully” and “The test has been completed. The results are as follows:”. It will then display a summary of the test results.

• Uploading Results: The application will upload the test results to the Fairly AI Platform and display a message with the URL where you can view the results.