This report analyzes the Boston Housing dataset to identify key drivers of residential property prices using multiple regression techniques in R. Four models are built and compared: - Full Linear Model (R² = 0.733) - Reduced Linear Model (insignificant variables removed) - Quadratic Model (R² = 0.832) - Log-Quadratic Model (best diagnostic behavior) Key findings: - Room count (rm) is the strongest positive predictor - Poverty rate (lstat) is the strongest negative predictor - Air quality (nox) significantly depresses home values - Quadratic model achieves lowest RMSE of 4.49 Tools used: R, ggplot2, broom, knitr, Quarto

This report forecasts US GDP growth rate using 64 years of quarterly Federal Reserve data (1960-2023). Two models are built and compared: - ARIMA — captures autocorrelation structure - ETS — adapts to structural shifts post-COVID Key findings: - GDP growth is mean-reverting (~2.5% long run average) - ARIMA performs better in stable economic periods - ETS adapts faster during volatile periods - Both models significantly outperform naive forecasts Tools used: R, forecast, tseries, Quarto