CLASSIFICATION AND REGRESSION TREES

STAT 451

Pattern classification, regression analysis, clustering, and dimensionality reduction. For each category, covers fundamental algorithms and selections of contemporary, current state-of-the-art algorithms. Focus on evaluation of machine learning models using statistical methods. Statistical pattern classification approaches, including maximum likelihood estimation and Bayesian decision theory, algorithmic and nonparametric approaches. Practical use of machine learning algorithms using open source libraries from the Python programming ecosystem.

STAT 841

Statistical function estimation and classification; reproducing kernel machines, support vector machines; high dimensional model selection and estimation; Bayesian, empirical Bayesian interpretation of nonparametric learning methods; log density ANOVA and graphical models; tree ensemble methods including bagging, boosting, and random forest.

STAT 761

Tree construction, including finding splits, tree-pruning and error estimation. Categorical predictor variables, missing or censored data, prior class-probabilities, and unequal misclassification costs. Selection bias. Comparison with other statistics and machine-learning methods. Extensions to piecewise linear and non-least squares regression models.

STAT 333

An introduction to regression with emphasis on the practical aspects. Topics include: straight-line model, role of assumptions, residual analysis, transformations, multiple regression (with some use of matrix notation), multicollinearity, subset selection, and a brief introduction to mixed models.

STAT 421

Analysis of multidimensional contingency tables, Poisson regression, and logistic regression, with emphasis on practical applications. Use of computer programs for such analyses. Model selection, testing goodness of fit, estimation of parameters, measures of association and methods for detecting sources of significance.