This project consists a semi-automated data science pipeline and explainable workflow for wind power forecasts using Lale & AIX360.
This workflow enables data scientists to train, tweak and validate the pipeline with better explainability. In production, the semi-automated nature of the workflow allows data scientists to take real-time weather and power data to quickly tune and re-train models.
Finally, because we use public data, anyone interested can develop their own models to forecast wind power with this robust workflow.
The work is primarily done on Watson Studio notebook running python 3.6.9. A requirements.txt for the dev environment is provided if you'd like to replicate the results.
