What data and approach for a simple 30-day DAU forecast?
Tests forecasting from sessionized logs without overengineering. Cite timestamped events, a 30 min session rule, and a regression baseline with day-of-week, recent totals, scored with MAE. Red flag: deep learning before a baseline or ignoring privacy hashing.
Tests whether you can build a practical daily forecast from raw event logs without over-engineering. A strong answer identifies time-stamped click, page load, and video play events; sessionizes them with a 30-minute inactivity threshold; hashes user IDs for GDPR/CCPA compliance; and constructs temporal features like day-of-week and hour. The modeling approach should be a simple regression baseline predicting daily totals, evaluated with MAE or RMSE, and deployed in batch with an A/B test plan.
Read the original → aijourn.com
- #growth
- #forecasting
- #feature-engineering
- #privacy
- #experimentation
Get five bites like this every day.
Tezvyn delivers a daily feed of 60-second tech bites with quizzes to lock in what you learn.