Scope: end-to-end time series pipeline to forecast flight passenger demand.
Data: mocked dataset with dates, routes, and passenger counts simulating airline traffic.
Feature engineering: vectorized time-based features (1 and 7 day lags + 7 day rolling average).
Temporal Split: strict chronological 80/20 train/val split (data leakage prevention).
MLflow Tracking: log hyperparameters, RMSE metric and the binary model artifact directly into mlflow.db (SQLite).
MVP: tracking experiments with MLFlow, comparing different time-series models + hyperparameter tuning.