Event Details
This talk presents a novel latent Markov model with Plackett-Luce emission distributions for the analysis of longitudinal ranking data. The proposed framework models preference dynamics through a finite number of latent preference regimes, allowing for abrupt changes in individual ranking behavior rather than assuming smooth temporal evolution. Individuals transition between latent states over time according to a first-order Markov process, capturing unobserved heterogeneity in preference trajectories. Model parameters are estimated by maximum likelihood via an Expectation-Maximization (EM) algorithm. An application to experimental task scheduling data illustrates the ability of the proposed model to identify distinct behavioral profiles, characterize temporal heterogeneity, and detect individual transitions between latent preference regimes.
