dotnet / machinelearning

ML.NET is an open source and cross-platform machine learning framework for .NET.
https://dot.net/ml
MIT License
8.92k stars 1.86k forks source link

Sample Project does not trigger anomalies with v3 #6994

Closed mjanulaitis closed 4 months ago

mjanulaitis commented 4 months ago

System Information:

Describe the bug I have successfully downloaded Microsoft's sample projects and run the AnomalyDetection_PhoneCalls sample with one change. That is, I upgraded from .Net Core 3.1 to .Net 7. Then after learning 3.x was released, I upgraded to the latest NuGet Package. Now when I run the sample it successfully runs however no anomalies are detected.

To Reproduce Steps to reproduce the behavior:

  1. Download the samples from here: https://github.com/dotnet/machinelearning-samples/
  2. Open the AnomalyDetection_PhoneCalls sample.
  3. Modify the project file to use net7.0
  4. Run. Verify several items are flagged as anomalies.
  5. Using NuGet Package Manager upgrade Microsoft.ML and Microsoft.ML.TimeSeries to v3.0.1
  6. Run. Notice the same items are no longer flagged as anomalies.

Expected behavior The same items should be flagged as anomalies.

mjanulaitis commented 4 months ago

I was able to resolve the issue by changing the default threshold and sensitivity.