Get startedGet started for free

Alerting

As you have heard in this lesson, you can monitor data in production by using the following:

  • deterministic methods: by checking if all required attributes are present and have values within strictly defined sets; and
  • statistical methods: by checking if the distribution of data observed in production is significantly different than the distribution observed at training time.

You have also heard that the latter methods can be very sensitive to even the smallest changes and generate a large number of uninformative (non-actionable) alerts.

But why is that actually an issue?

This exercise is part of the course

MLOps Deployment and Life Cycling

View Course

Hands-on interactive exercise

Turn theory into action with one of our interactive exercises

Start Exercise