Continuously track models in production to automatically detect drift and maintain accuracy in rapidly changing live environments
Nothing lasts forever—not even carefully constructed models that have been trained using mountains of well-labeled data. In these turbulent times of massive global change emerging from the COVID-19 crisis, ML teams need to react quickly to adapt to constantly changing patterns in real-world data. Monitoring machine learning models is a core component of MLOps to keep deployed models current and predicting with the utmost accuracy, and to ensure they deliver value long-term.
Machine learning model monitoring is a critical element of the production environment. When models encounter data that is significantly different from the training data—due either to limitations in the training data or from changes in the live environment—previous data becomes obsolete. The Iguazio Data Science Platform detects these occurrences on the feature level, monitors the model outcome, creates an alert when performance degrades and then triggers automatic retraining. The entire data pipeline is fully managed and monitored on Iguazio.
Machine learning model monitoring is natively built in to the Iguazio Data Science Platform, along with a wide range of model management features and ML monitoring reports. Monitor all of your models in a single simple dashboard.
Automatically detect concept drift, anomalies, data skew, and model drift in real-time. Even if you are running hundreds of models simultaneously, you can be sure to spot and remediate the one that has drifted.
When drift is detected, Iguazio automatically starts the entire training pipeline to retrain the model, including all relevant steps in the pipeline. The output is a production-ready challenger model, ready to be deployed. This keeps your models up to date, automatically.
Feature vectors and labels are stored and analyzed in the Iguazio feature store and are easily compared to the trained features and labels running as part of the model development phase, making it easier for data science teams to collaborate and maintain consistency between AI projects.
