Why is my newly created trainable classifier in Microsoft Purview only showing 5 test results in the "Review test results" tab of the classifier when I have 100 positive and 277 negative sample files in SharePoint Online that I used to train the model on?

Question

Why is my newly created trainable classifier in Microsoft Purview only showing 5 test results in the "Review test results" tab of the classifier when I have 100 positive and 277 negative sample files in SharePoint Online that I used to train the model on?

Please advise, thank you.

Answer 1

Hi @Josh Wilson

Greetings & Welcome to Microsoft Q&A forum! Thanks for posting your query!

The limited number of test results displayed in the "Review test results" tab of your newly created trainable classifier in Microsoft Purview could be due to several factors such as:

Processing Time - After creating a trainable classifier, it processes the seed data and builds a prediction model, which can take up to 24 hours or less. During this time, the classifier status will show as "In progress." Once processing is complete, it should provide more comprehensive results.

Seed Data Indexing - If the classifier is still in the early stages of evaluation or if there are issues with the seed data being indexed properly, this could also affect the number of results shown. Ensure that the seed data is correctly set up and that you allow sufficient time for the classifier to process all the samples.

Automatic Testing - The improved version of the trainable classifier feature now includes automatic testing, which means that a test set of both positive and negative samples is no longer required at the testing stage. This might also influence the number of test results displayed.

If you continue to experience issues, it might be helpful to review the setup and ensure that all steps have been followed correctly.

For additional information, please refer: How to create a trainable classifier

I hope this information helps, please do let us know if you have any further queries.

Thank you.