AnDOUC TechCast 2025-05-01 - Data & Model Monitoring - The Step Not To Skip In Solution Deployment скачать в хорошем качестве

AnDOUC TechCast 2025-05-01 - Data & Model Monitoring - The Step Not To Skip In Solution Deployment

Presented by Mark Hornick, Oracle Often, the final step in the machine learning solution deployment step goes unfinished – monitoring. Data monitoring helps ensure data integrity for your applications and dashboards by identifying data drift and gaining insight into data changes. Model monitoring helps identify changes in model accuracy which can result from, e.g., changes in customer behavior as well as upstream data quality. Detect issues early and take corrective action before these issues have a significant impact. In this session, explore the monitoring problem, where it fits in the ML process, and how to use no-code data and model monitoring user interfaces in Autonomous Database. Join us to learn about this key MLOps topic along with demonstrations. ---------- Chapters 00:00 Introduction 05:35 ML Monitoring: Canary in the Coalmine 09:40 CRISP-DM 14:20 Data Monitoring 22:15 Model Monitoring 32:40 Data Bias Detection 37:15 Taking Informed Action 39:00 Q+A 42:00 Wrap-Up ----------