Courses: SYSC 610: Discrete Multivariate Modeling-II -- Spring 2007

From description of SySc 651 (which is DMM-I): In this course, information theory is used as a framework for modeling and data mining: for analyzing static or dynamic relations among discrete variables, for detecting complex interaction effects, and for discovering nonlinearities in continuous variables made discrete by binning. In the systems literature, these information-theoretic and related set-theoretic methods, used together with graph theory techniques, are called "Reconstructability Analysis" (RA). RA overlaps with and extends log-linear modeling in the social sciences, Bayesian networks and graphical models in machine learning, decomposition techniques in multi-valued logic design, Fourier methods for compression, and other modeling approaches. It can be used for confirmatory and exploratory statistical modeling as well as for non-statistical applications.

SySc 610 continues the presentation of discrete multivariate modeling (SySc 551/651), and will focus on (a) projects and (b) advanced topics. In projects, students will either do an intensive analysis of some dataset or a software project that enhances the current set of RA tools. The advanced topics will include:

• state-based RA and k-systems analysis
• RA loopless models with many variables ("dependency analysis")
• identification with inconsistent data
• binary decision diagrams and set-theoretic RA
• intra-model analyses
• modeling with latent variables
• RA and genetic algorithms
• Fourier-based RA techniques
• the OCCAM software package
• binning

Prerequisites: SySc 651 (or -- with permission of the instructor -- very solid background in log-linear modeling, Bayesian networks, or related methods)

TEXT: A reader available at SmartCopy (in addition to the texts used for SySc 551/651). Grades will be based on a major data analysis project.

Grades will be based on a major data analysis project.

Spring 2007 Syllabus