Life, Death, and Shopping
summary
Covers using survival-analysis framing for events like churn, engine failure, or purchases, highlighting that labels are often rule-derived (e.g., 30-day inactivity) and that the data layout differs from standard ML. Explains how to prepare and aggregate time-based features with PySpark for these problems, and describes a connection to NLP techniques for this kind of temporal/event data.
AI-written summary · about
related talks
12mLife, Death, and Shopping
35mLife, Death, and Shopping
53mMachine Learning in E‑commerce
44mEating Humble Py: From Toy Problem to Real-World Solution in Predicting CLV
17mCreating & operating ML models from event-based data using feature stor...
36mCustomer Lifetime Value: Models, Metrics and a Multitude of Uses
31mLife cycle of data becoming a model
49mUnderstanding Customer Journeys with Deep Learning