Attendance Forecasting & Prioritisation

What this is

Benchmarked SARIMA, Prophet, Holt-Winters and Random Forest across six categories on out-of-sample MAE; a tuned Random Forest roughly halved error on the hardest categories. Paired with ABC/Pareto prioritisation. Genuine model-selection rigour, not optimisation.

Metrics

Stack

What I would not claim

Nothing beyond what is listed above.