DataRobot has an inhouse algorithm named XEMP for prediction explanation. Unfortunately they haven't released the algorithm which makes it difficult to trust and use it. There is a white paper which compares it with LIME. That document is more a product marketing document than a technical document.
Does anyone have any technical understanding of how this algorithm works and what is the experience of people using it?
eXemplar-based Explanations of Model Predictions
It works well with predictions explanation
Pls see the DataRobot docs here -https://docs.datarobot.com/en/docs/modeling/analyze-models/understand/pred-explain/predex-overview.html
And here is a great Stanford page that chrome-extension://bdfcnmeidppjeaggnmidamkiddifkdib/viewer.html?file=https://hci.stanford.edu/courses/cs147/2019/au/readings/example_based_explanations_ml_interface.pdf