Effortless model deployment with MLflow — Stratified models (many models) for forecasting

How to package stratified models or partitioned models into a single MLflow entity that you can seamlessly deploy to produce predictions.

Welcome back to the series Effortless model deployment with MLflow! If you just join the party, check out the other post of the series:

• MLFlow: Introduction to the MLModel specification.
• Customizing inference with MLFlow: deploying a Computer Vision model with fast.ai.
• Packaging models with multiple pieces: deploying a recommender system.
• Packaging models with multiple assets.
• Packaging many models: deploying a partitioned model for demand forecasting (this post)

In my previous post from the series, Effortless models deployment with Mlflow — Packaging models with multiple assets, we saw how we can package a model that is composed of multiple pieces using MLflow. Those pieces were for instance data transformations, feature encoders, and… of course, your model. But can those pieces be “multiple models”?

Sometimes when we are modeling a given problem, we recognize that there are sections of the…