CS 652 Algorithms for Statistical Learning

The course introduces modern methods for statistical learning (prediction). Students learn how to apply these methods using the programming language R (or Python). In particular, the course includes: (1) the subject of statistical learning, prediction accuracy and model interpretability, supervised and unsupervised learning, regression vs. classification; (2) model accuracy, measuring the quality of fit, bias-variance trade-off; (3) linear regression and its extensions, multiple linear regression, qualitative predictors; (4) resampling methods, cross-validation, bootstrap; (5) linear model selection and regularization: subset selection, shrinkage methods (in particular, ridge and lasso), dimension reduction methods; (6) non-linear models, splines; (7) tree-based methods, decision trees, random forests, boosting; (8) support vector machines, maximal margin and support vector classifiers.

Credits

3

Prerequisite

CS 650 Introduction to Big Data