STAT 5608: Applied Modern Multivariate Statistical Learning
Credits: 3. Contact Hours: Lecture 3.
Prereq: STAT 5501, STAT 5542, and STAT 5579
A Statistics-MS-level introduction to Modern Multivariate Statistical Learning. Theory-based methods for modern data mining and machine learning, inference and prediction. Variance-bias trade-offs and choice of predictors; linear methods of prediction; basis expansions; smoothing, regularization, kernel smoothing methods; neural networks and radial basis function networks; bootstrapping, model averaging, and stacking; linear and quadratic methods of classification; support vector machines; trees and random forests; boosting; prototype methods; unsupervised learning including clustering, principal components, and multi- dimensional scaling; kernel mechanics. Substantial use of R packages implementing these methods. Offered even-numbered years.
(Typically Offered: Spring)