This unit introduces the shift from business intelligence to data science and the evolution from CRISP-DM to modern, multicomponent machine learning systems, including the rise of foundation models and LLMs. Students will explore production principles, off‑the‑shelf components, and a data‑driven design approach through the example of a YouTube search agent to prepare for hands‑on model building and deployment in subsequent units.
Learning Objectives
- Analyze key differences between traditional business intelligence workflows and modern data science/ML pipelines, identifying implications for data needs, team roles, and outcomes
- Compare the CRISP-DM methodology with contemporary multicomponent ML system lifecycles and list at least three additional components or lifecycle activities (e.g., model serving, monitoring, data engineering)
- Evaluate how foundation models and large language models (LLMs) affect system complexity, capability, and reuse, and determine criteria for selecting off‑the‑shelf components versus custom models
- Design a component-level plan for a YouTube search agent that applies production ML principles and a data-driven approach, specifying data sources, model components, evaluation metrics, and monitoring/maintenance strategies
Curriculum
- 1 Section
- 9 Lessons
- 30 hours
Expand all sectionsCollapse all sections
- Unit 1 — Foundations of Modern ML Systems9
- 1.1Data Science Is Diverging from Business Intelligence20 mins
- 1.2From CRISP-DM to Modern, Multicomponent ML Systems20 mins
- 1.3The Emergence of LLMs Has Increased ML’s Power and Complexity20 mins
- 1.4What You Can Expect from This Book20 mins
- 1.5Components of a YouTube Search Agent20 mins
- 1.6Principles of a Production Machine Learning System20 mins
- 1.7The Emergence of Foundation Models20 mins
- 1.8The Role of Off-the-Shelf Components20 mins
- 1.9The Data-Driven Approach20 mins
Instructor
Free
Leave a Reply