Automating the process of continuous training and evaluation of machine learning models on Databricks using MLflow requires a systematic and efficient approach. Scheduling a Databricks job to run a notebook that uses MLflow for tracking experiments is the most suitable method for several reasons:

1. Automation: This approach automates the process of retraining and evaluating models weekly, ensuring models are consistently updated without manual intervention.
2. MLflow Integration: MLflow is designed for managing the machine learning lifecycle, making it ideal for tracking experiments and monitoring model performance.
3. Scalability: Databricks provides a scalable environment for machine learning workloads, allowing for efficient handling of large datasets and complex models.
4. Consistency: A weekly schedule ensures models are regularly updated with the latest data, maintaining their accuracy and relevance over time.

In summary, scheduling a Databricks job to run a notebook with MLflow integration is the most efficient way to automate continuous training and evaluation, leveraging Databricks' capabilities and MLflow's tracking for systematic model improvement.