Goal

• Predict the duration of a taxi trip.

Steps

1. Design: Is the ML suitable for solving our problem?
2. Train: We have an ML model, and we train it with our data.
3. Operate: It includes deploying the model on a web server with an API to be used by the user to solve the problem.

Configure the environment with GitHub Codespaces

1. Create a new GitHub repo with a README.md file and a .gitignore file with a Python template.

2. Click on Code, then Codespaces, and finally create codespace on main.

3. Open the codespace on VS Code.

4. Download and install the Anaconda distribution of Python.

5. Run jupyter notebook to start the Jupyter server.

6. For reference:

   https://github.com/DataTalksClub/mlops-zoomcamp/tree/main/01-intro

MLOps Maturity Model

• We should be pragmatic, as it’s ok to start with level zero and up your level to level 2 with model monitoring, as not all ML models need to be in level 3 and 4.

Level 0

• No automation, neither MLOps
• We do everything using a sloppy Jupyter notebook without automation or ML pipelines.

Level 1

• DevOps, but No MLOps