In my Kaggle notebook, I used UCI's simulated Electrical Grid Stability data to predict if a given combination of power system conditions would result in an unstable grid - and therefore risk causing blackouts or damaging equipment.
I then created a simple streamlit app that can be used to see how adjusting model inputs affect the resulting predictions:
This repo contains the source code as well as the associated tests and github actions.
The notebook goes through data exploration and feature selection, followed by model fitting, tuning, testing and pickling. Here are some images for an idea:
