STAT 558

Statistical Machine Learning for Data Scientists

Course Description

Bias-variance trade-off; training versus test error; overfitting; cross-validation; subset selection methods; regularized approaches for linear/logistic regression: ridge and lasso; non-parametric regression: trees, bagging, random forests; local regression and splines; generalized additive models; support vector machines; k-means and hierarchical clustering; principal components analysis. Prerequisite: STAT/BIOST/DATA 557, or permission of instructor. Offered: jointly with BIOST 558/DATA 558; Sp.