sachn1/iml-lime

RUSKLAINER - identification of features that contribute towards RUpture riSK prediction in intracranial Aneurysms using Lime explAINER

This work is done to explore LIME surrogate model on a dataset. We start by deep exploration and analysis of data which we then preprocess and apply several black-box models on it. Then we build a LIME surrogate model on the selected best black-box model. Additionally, the majority of our work is done in:

1. Understanding and testing the effects of each hyper-parameter of LIME and see its limitations.
2. Building an on-the-cloud GUI application, which serves two purposes; Allows IML (Interpretable Machine Learning) user to see the effect of each hyperparameter on individual features and see the working of LIME in real-time. Secondly, allowing subject matter expert to view the findings and derive inferences from it.

Feel free to explore the real-time interactive tool RUSKLAINER!