STSCI 4600 - Statistics for Risk Modeling

Fall. 3 credits. Student option grading.

Prerequisite: STSCI 3080. Prerequisite or corequisite: STSCI 4030. Not offered every year.

J. Entner.

This course will provide an introduction to methods and models for analyzing data. Methods for selecting and validating models with actuarial applications will be emphasized. Topics to be covered will include regression models (including the generalized linear model), time series models, principal components analysis, decision trees, and cluster analysis.

Outcome 1: Explain the types of modeling problems and methods, including supervised versus unsupervised learning and regression versus classification and the common methods of assessing model accuracy.

Outcome 2: Employ basic methods of exploratory data analysis, including data checking and validation.

Outcome 3: Estimate parameters using least squares and maximum likelihood.

Outcome 4: Interpret diagnostic tests of model fit and assumption checking, using both graphical and quantitative methods.

Outcome 5: Calculate and interpret predicted values, confidence, and prediction intervals.

Outcome 6: Interpret the results of a principal components analysis, considering loading factors and proportion of variance explained.

Outcome 7: Explain the purpose and uses of decision trees.

Outcome 8: Explain and interpret decision trees, considering regression trees and recursive binary splitting, bagging, boosting, random forests, classification trees, their construction, Gini index, and entropy.

Outcome 9: Interpret the results of a decision tree analysis.

Outcome 10: Explain K-means & hierarchical clustering.