About fault to selection ratio

[FasihMunirMalik]

The fault-to-selection ratio being used as an evaluation metric returns:
return nr_detected_faults / len(test_suite)
where as it should be nr_detected_faults / len(selection)

so incase of a test suite of 192 cases where I select 74, 51 out of these 74 fail so fault to selection ratio should be 51/74, currently It is 51/192

[ChristianBirchler]

Thanks @FasihMunirMalik ! It'll be fixed (#27 ).

[FasihMunirMalik]

@ChristianBirchler
There is a missing package in Docker image, I was unable to assign this as an issue to you
Description:
A recent update introduced the use of scikit-learn in the evaluation.py file, specifically functions like pairwise_distances from sklearn.metrics.pairwise. However, scikit-learn is not currently installed in the Evaluator Docker image, which leads to an import error when running the container.

Steps to Reproduce:

1. Build latest Docker image.
2. Run the container.
3. Observe the import error for scikit-learn.

Current Behavior:
Running the Docker container results in an error indicating that scikit-learn is missing.

Expected Behavior:
The Docker image should have scikit-learn installed so that functions like pairwise_distances work without errors.

Suggested Fix:
Add scikit-learn to the Evaluator Dockerfile. For example, you could include:

RUN pip install scikit-learn

This would ensure that scikit-learn is available in the container, allowing for successful execution of scripts that depend on it.

Thank you for addressing this issue!