Developed time series forecasting models on hourly streaming viewership data using
Simple, Double, and Triple Exponential Smoothing, as well as
SARIMA with seasonal differencing and bootstrapped confidence intervals.
Conducted EDA, trend/seasonality decomposition, and calendar-effect analysis; optimized forecasting
accuracy via solver tuning and validation on a 75/25 time-based split, minimizing
Mean Absolute Error (MAE).