This report highlights different methodologies used in Time Series Forecasting by using the average monthly births in New York City (1946 - 1959) to help forecast the city's growth. I have used Polynomial Regression, Regularized Regression (LASSO, Ridge, Elastic Net), Holt-Winters and Box-Jenkins (SARIMA) models. The best model is selected based on Residual Diagnostics and Average Predictive Error (APE) to forecast the average births in NY between 1960 - 1961.
It serves as a good introduction to Time Series Data modelling and ways to understand models that provide a better fit versus stronger predictive power.