Graphical Modeling

Credits: 
2.0
Term: 
Fall
Course Description: 

Advanced Methods course, open for ALL tracks

Cancelled in Fall 2017/2018

The course gives an introduction to the theory of many of the currently applied graphical models. All these models attempt to describe the associations and effects among the variables, and some may be given causal interpretation. The graphical nature of these models means that the structure formulated in the models may be illustrated by a graph, and graph-related language may be used to describe and interpret the meanings of the models. The models apply equally to discrete and continuous data. Several classical models used in multivariate statistics nay be formulated as a graphical model, with additional insight gained using graphs. The emphasis will be on the structural aspects, so the material is relevant, whether the data are from a survey, from administrative records, or are 'big'. This is an advanced class, assuming the students are familiar with statistical analysis and are able to explore and use statistical software (mostly in R) on their own. Much of the material presented is necessary background to study some of the contemporary theories of causality.

Learning Outcomes: 

Ability to critically choose, apply and interpret graphical models in a wide range of statistical problems, including variables of different levels of measurement, from experiments or surveys. Answer research questions regarding associations and effects among variables, including exploratory and confirmatory analyses and causal modeling.

Assessment: 

Midterm test – 30%
Final take-home assignment – 70%
The assignment will require students to analyze data of their own choice using methods covered in class.
Students who wish to audit the class will have to complete the midterm test