Solving Real Problems with Data Science

Description from DukeHub

The hardest part of most data science classes is often model fitting, but the hardest part of being an effective data science professional is understanding your stakeholder's problem, fitting the right model for the problem being solved, and communicating findings. This class fills the gap between neatly curated classroom exercises and the messiness of the real world. It provides a unified perspective on how the material learned in other courses complements one another, and when different approaches to data science are most appropriate to use. It provides a framework for understanding your goals as a data scientist, and how to achieve them. Assumes familiarity with introductory statistical inference and machine learning, covers but does not assume familiarity with causal inference.

Skills

• data and statisticsdescription: "data science" and 4 more

AI-resilient: 0 of 5 0/5

• No: Physical, lab or field work
• No: Live work with people
• No: Judgment someone is accountable for
• No: A craft you practice with your hands or body
• No: Small, in-person format

The checks lean on Microsoft Research, Working with AI (2025), which found generative AI applies least to physical work and hands-on work with people. A signal, not a promise. How the score works

More in IDS

• Design Your Own Social and Behavioral Research IDS 306S
• Practical Data Science II IDS 541
• Applied Machine Learning IDS 542
• Introduction to Data Science and Quantitative Methods in Risk Management IDS 583
• Data Science and Decision Optimization in Banking & Financial Services IDS 586
• Credit Risk Modeling IDS 598-01

All IDS courses · All of Graduate School