ACT2039/Employee-Attrition-Prediction ? reverse-engineered prompt

Build me an end to end employee attrition prediction project in Python that I can run in Google Colab or Jupyter Notebook.

Use the IBM HR attrition dataset to explore what drives people to leave, clean and prepare the data, do clear EDA with charts, and train a few classification models like Logistic Regression, Random Forest, and Gradient Boosting. I want the notebook to compare the models with accuracy, precision, recall, F1 score, and ROC AUC, then show the best one clearly.

Please include useful visualizations for attrition by department, job role, income, work life balance, years at company, and a correlation heatmap. Also show feature importance and give a simple business summary with practical HR recommendations based on the findings.

Make it polished and easy to follow, with everything needed to run the notebook from start to finish. If you need current library details or best practices, look them up online first.