Life of a data scientist
summary
Claims that most time goes to data prep not building AI, follows CRISP-DM, and modeling is often the easiest step — sometimes automated to produce 42 models at once. Gives concrete examples: an IKEA footfall predictor that turns out to be driven mainly by weather and a working festival recommender, and an insurance fraud use case about Ray‑Ban claims. Stresses monitoring for model drift and that deployment and ethics/explainability determine whether a model actually works.
AI-written summary · about
chapters (8)
- 0:00Start
- 12:03Understand the business
- 13:23Collect and combine the data
- 16:43Data preparations
- 20:00Modeling: the easy part
- 22:26Pick the "best" model
- 26:40Explain to the business
- 28:07Production time
related talks
56mData, AI, Hype and Fear - Recipe for Disaster?
43mLearning the fundamental elements to build effective AI/ML products
28mAI/ML for Non Data Scientists
16mOperationalizing AI in Finance: Real-Time Analytics
31mLife cycle of data becoming a model
46mThe AI dominated world – skills required for successfully navigating it
19mData science demystified
25mWhy Good Models Fail After Deployment