prio-data / views_pipeline

VIEWS forecasting pipeline for monthly prediction runs. Includes MLops and QA for all models/ensembles.
Other
3 stars 3 forks source link

Issue: Detailed Review of PR #96 – Logging Strategy, Storage Decisions, and Future-Proofing #108

Open Polichinel opened 2 weeks ago

Polichinel commented 2 weeks ago

Issue: Detailed Review of PR #96 – Logging Strategy, Storage Decisions, and Future-Proofing

Description
Conduct an in-depth review of PR #96, focusing on the proposed logging strategy, storage locations, and logger levels. This review will help establish best practices for logging across the project and ensure future scalability. Additionally, verify that new loggers integrate consistently with existing logging functions.

We will update ADR 015 accordingly.

Review Objectives

  1. Assess Logging Levels and Storage Decisions:

    • Evaluate the choices for logging levels. Consider whether the levels used are too general or if a domain-specific approach would enhance clarity. "Error" for instance is very general whereas "model_training_failed" or something like that is more domain-specific
    • Review storage locations for logs and help decide if the current setup is optimal or if alternative storage should be considered for future scalability.
    • Identify any areas where current storage or level selection may limit flexibility or cause bottlenecks for future development.
    • Refer to questions noted in the PR and ensure recommendations are backed by MLOps best practices.
  2. Create and Test Logging Functionality:

    • Test if existing logging functions are compatible with the new implementation.
    • Consider developing unit tests that validate the logging setup to support ongoing and future testing. Ensure unit tests follow a standardized naming convention to support future maintenance and reuse.
  3. Future-Proofing Recommendations:

    • Help refine the current approach to make it more sustainable and scalable, especially if model or dataset complexity increases. Clarify any points necessary to ensure alignment with Simon’s goals for the PR, aiming to make the logging strategy flexible and adaptable.
    • Suggest improvements or provide feedback that could help structure an ADR capturing these logging decisions.

Tasks

Next Steps
Once the review is complete, simon will update the ADR to outline our logging strategy and practices for consistency across the project, supporting both current needs and future scalability. The next step after that is updating the scripts to correspond to the ADR

Labels
review, logging, PR review, MLOps, future-proofing