Retraining a machine learning model

Concetti di MLOps

Folkert Stijnman

ML Engineer

Retraining after changes

 

Retraining: use new data to develop a fresh version of the machine learning model

Drift in data

Data drift

Data drift: changes in the input data

Concept drift

Concept drift: changes in the relationship between input and output data

How often to retrain?

• Business environment: how volatile is the data?
• Cost: how much does it cost to retrain?
• Business requirements: what is the required model performance?

Retraining methods

Retraining methods

Automatic retraining

Let's practice!

Concetti di MLOps