Prof. John K. Kruschke
Brief Description: This seminar is intended as a follow-up to the version of Q550 Prof. Kruschke taught in Spring 2003. Other students with previous experience modeling data are also welcome. ¶ In this course we will explore the mechanics of fitting models to data, and we will delve into theoretical issues in deciding which models fit best. We will consider topics such as different measures of discrepancy between data and model predictions, different algorithms for finding best-fitting parameter values, estimating confidence intervals for parameter values, detecting and dealing with parameter redundancy, measures of model complexity, and criteria for model selection. ¶ The material will be discussed mostly at a general, methodological level, with only occasional application to specific models in cognitive science. Lots of "hands on" experience will be gained via work with MATLAB. See the schedule page for more details about topics and lots of cool pictures.
Required Textbook: Morgan, B. J. T. (2000). Applied Stochastic Modelling. New York: Oxford University Press.
Other Readings: Recent articles available online to IU students. Among these will be articles by In-Jae Myung regarding model complexity. Details TBA.
Software: We will make extensive use of MATLAB. MATLAB is available on all IU public cluster computers. Details of campus availability can be found from the Stat-Math Center. MATLAB is available for purchase in student versions, see its manufacturer, MathWorks.
A free Matlab-alike, called Octave, is also available: http://www.octave.org/. (Thanks to Christoph Weidemann for informing me of Octave.)
Registration Info: Course number P747, section 3885. Requires graduate standing in a field related to cognitive science or permission of the instructor.
Time and Place: Spring semester, 2004. Mondays and Wednesdays, 10:10-11:25am, Psychology Building Room 113.
This web page is at URL= http://www.indiana.edu/~jkkteach/P747_2004/