|Time:||12:30 pm on Wednesday, March 7, 2012|
In the US Census Classification (1970) jobs are pairs of 3-digit (Industry x Occupation) codes. Any analysis of career data and dynamics of employment has to cope with the inherent high dimensionality of this representation. We work under the assumption that "job" is actually a continuous variable, and infer the continuity from the transitions people take between jobs, as recorded by the National Longitudinal Survey of Youth (NLSY). Thus, proximity between jobs will mean that they are “close” to each other in a non-negligible subset of career paths. The embedding we obtain allows one to visualize the job landscape. Moreover, one can map individual or groups of career paths to this space, extract features of their collective structure, and construct statistical tests comparing groups by means of this mapping.
On the theoretical side, we show that this work is an instance of fitting a particular generative model for asymmetric relational data. We provide theorems that ground and motivate our algorithms, and that ensure asymptotic consistency of the results.
Joint work with Dominique Perrault-Joncas (UW Statistics) and Marc Scott (NYU)
To request disability accommodations, contact the Office of the ADA Coordinator ten days in advance of the event. 543-6450 (voice) 685-3885 (FAX) 543-6452 (TDD) firstname.lastname@example.org