This report explores key statistical learning methods including ridge regression, lasso, principal components regression (PCR), partial least squares (PLS), and best subset selection. The analysis examines the trade-offs between model complexity, interpretability, and prediction accuracy. Using the College and Boston datasets from the ISLR2 package, various models are evaluated using validation set and cross-validation test errors. The report emphasizes model selection, performance comparison, and the interpretability of regression models in predicting college applications and crime rates.

This assignment applies statistical learning techniques to the Default and Boston datasets from the ISLR2 package. It explores k-fold cross-validation and validation set approaches to estimate test error rates. The bootstrap method is used to estimate standard errors for model coefficients and population parameters such as the mean, median, and 10th percentile of medv in the Boston data.

This report presents a series of classification analyses using the Boston, Weekly, and Auto datasets from the ISLR and ISLR2 packages. Logistic regression, linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), Naive Bayes, and k-nearest neighbors (KNN) were applied to predict key binary outcomes: whether a Boston census tract has a crime rate above or below the median, the market direction in the Weekly dataset, and whether a car's fuel efficiency is above or below the median in the Auto dataset. The results include confusion matrices, error rates, and key findings regarding model performance.

This report explores multiple linear regression techniques using the Auto and Carseats datasets from the ISLR package. The analysis includes correlation assessment, model fitting, interpretation of coefficients, diagnostic plots, interaction terms, and variable transformations to improve model performance.