Methods of Data Science

Course Description

A dramatic increase in computing power has enabled new areas of data science to develop in statistical modeling and analysis. These areas cover predictive and descriptive learning and bridge between ideas and theory in statistics, computer science, and artificial intelligence. We will cover methods including predictive learning: estimating models from data to predict future outcomes. Regression topics include linear regression with recent advances using large numbers of variables, smoothing techniques, additive models, and local regression. Classification topics include linear regression, regularization, logistic regression, discriminant analysis, splines, support vector machines, generalized additive models, naive Bayes, mixture models and nearest neighbor methods as time permits. Lastly we develop neural networks and deep learning techniques, bridging the theory introduced in the earlier parts of the class to purely empirical methods. We situate the course components in the "data science lifecycle" as part of the larger set of practices in the discovery and communication of scientific findings. Prerequisite: IS 507 or equivalent (e.g. intro probability/stats STAT 100, CS 361, or ECON 202); and IS 490 IDS/CS 398 ID/STAT 490 or CS101 or equivalent; or consent of the instructor. Linear Algebra recommended at the level of MATH 125; Calculus recommended at the level of MATH 220.

Code

IS 517

Credits

4

Subject

Information Sciences

Programs That Include This Course

• Doctoral Doctor of Philosophy in Informatics
• Specialization Information Sciences Concentration (Graduate)