Which of the following items should be included in a handover to the end user to enable them to use and run a trained model on their own system? (Select three.)

A successful model handover ensures the end user can independently and correctly run the model. This requires three key components: the codebase itself, typically shared via a version-controlled repository (GitHub); comprehensive documentation (a README file) explaining dependencies, setup, and execution steps; and sample input/output files. The sample files allow the user to verify their setup is correct and understand the expected data formats. Together, these items make the model usable, understandable, and verifiable for the end user.

A. The developer's local folder structure is irrelevant to the end user and is not a portable or professional way to structure a project for handover.
B. Intermediate data files are artifacts of the training process and are not necessary for running inference with the final, trained model. They add unnecessary clutter.

1. Google. (n.d.). Rules of Machine Learning: Best Practices for ML Engineering. Rule #38: "Don't be a hero. Document your model." Mountain View, CA: Google. (This guide emphasizes the critical importance of documentation and providing all necessary artifacts for others to use the model).

2. Mitchell, M., et al. (2019). Model Cards for Model Reporting. Proceedings of the Conference on Fairness, Accountability, and Transparency, pp. 220-229. https://doi.org/10.1145/3287560.3287596. (The Model Card concept advocates for comprehensive documentation covering model usage, which aligns with providing code, instructions, and examples).

3. MLflow Documentation. (n.d.). MLflow Models. Retrieved from https://mlflow.org/docs/latest/models.html. (MLOps tools like MLflow package models with their code, dependencies, and an input example schema, reinforcing the importance of code, documentation, and sample data for reproducibility).