Husnain-Yaqoob/IT-Incident-Data-Analysis ? reverse-engineered prompt

Build me a Jupyter notebook project that takes IT incident event data and predicts how long each incident will take to resolve.

I want the notebook to load the incident CSV, clean and prepare the data, calculate resolution time, explore the main patterns with charts, and then train a few regression models to compare results. Please include a clear workflow from raw data to a final prediction model, and make sure the notebook explains what is happening in plain language as it goes.

Use the incident fields like priority, impact, urgency, category, reassignment count, reopen count, and system modification count to predict resolution time. Try a linear model, a decision tree, and a random forest, then pick the best one based on performance. Save the trained model and any label encoders so the prediction step can be reused.

If possible, add a simple Gradio interface where someone can enter the incident details and get back the estimated resolution time in hours, a rough risk level, and the estimated date and time it might be resolved.