Overview
What this challenge is about.
Build forecasts at 14-day horizon per region using: (1) classical baseline — SARIMA or Prophet; (2) ML approach — gradient-boosted regressor on engineered features (day-of-week, holiday flags, prior-window stats, content-drop indicators, marketing-spend lags). Use 18 months of historical DAU + event-calendar data (provided). Evaluate with proper rolling-origin cross-validation. Report MAPE per region + per horizon (day 1-14). Compare against the finance spreadsheet on held-out months. Build a Streamlit dashboard so live-ops can see forecasts + confidence intervals per region. Deliver: model code + tests, 6-page evaluation report, Streamlit dashboard, 3-page operationalization plan.
The Brief
What you'll do, and what you'll demonstrate.
Forecast DAU 14 days ahead per region with MAPE beating the finance-team spreadsheet baseline.
Earning criteria — what you'll demonstrate
- Apply both classical and ML approaches to time-series forecasting
- Use rolling-origin cross-validation for honest forecast evaluation
- Engineer time-aware features (lags, calendar effects, event indicators)
- Operationalize forecasts for live-ops decision-making
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Product Manager
PMs on live-ops products need forecasting fluency to time content drops and marketing pushes without burning revenue windows.
This challenge sharpens
- business-analytics
- time-series-forecasting
- model-evaluation