Describe the bug
In explanation.py when creating perf dictionaries for interpret show method actual_score and predicted_score need to be probabilities
To Reproduce
See review comments in PR #445 for details
Expected behavior
We need to get the probabilites and set them here for classification case