Which of the following is NOT typically included in the model evaluation process?

The model evaluation process focuses on assessing how well a model performs and understanding its effectiveness based on specific metrics and comparison to actual outcomes. The correct answer, which indicates what is typically not included in this process, is deployment of the model.

Deployment refers to the stage where the model is put into operation, often in a production environment for practical use. While deployment is a crucial part of the overall lifecycle of a data science project, it occurs after evaluation and is not directly about assessing the model's performance.

On the other hand, reviewing training data helps assess whether the model has had sufficient information to learn from and may uncover biases that could influence model performance. Comparing the model's predictions with actual outcomes is central to evaluating its accuracy and effectiveness. Lastly, checking assumptions and diagnostics is vital for understanding whether the model adheres to the underlying statistical requirements, ensuring that the applied techniques are valid.

Thus, while all other choices directly contribute to the model evaluation process, deployment is an action taken post-evaluation, making it the correct choice for what is typically not included in the evaluation process.