What Sadia demonstrated

Built an end-to-end demand forecasting workflow from 9,648 historical transactions: explored sales patterns, resampled to monthly revenue, decomposed trend and seasonality, and benchmarked moving average, linear regression, and SARIMA models with honest holdout metrics. The final comparison showed SARIMA as the strongest model, and the submission paired the technical work with executive-level takeaways on regional performance and channel mix.

• Leakage-Free Time-Series Splitting
• Seasonality and Trend Diagnosis
• Model Comparison with Error Metrics