Rankings, Inversions and Statistics

How do we do "statistics as usual" when data comes in the form ofpermutations, partial rankings, or other objects with richcombinatorial structure? I will describe how combinatorics,optimization and statistics come together in this endeavor. I willstart with a simple but very flexible family of models forpermutations. Then I will describe how this family can be model dataof various kinds (full rankings, partial rankings, signedpermutations). Some highlights will be results enabling practical Bayesiannonparameteric modeling over sets of partial rankings, and algorithmsfor discovering the structure of preferences in a population. Joint work with: Chris Meek, Harr Chen, Raman Arora, Alnur Ali, Bhushan Mandhani, Le Bao, Kapil Phadnis, Arthur Patterson, and Jeff Bilmes 12:30pm lunch, 1pm learner lecture, 3:30pm seminar, 4:30pm reception