A new spatial count data model with time-varying parameters

Recent crash frequency studies incorporate spatiotemporal correlations, but these have two key limitations – i) none of accounts for temporal variation in model parameters; and ii) Gibbs sampler suffers from convergence issues due to non-conjugacy. To address the first limitation, we propose a new count data that identifies underlying patterns regression parameters while simultaneously allowing time-varying spatial correlation. The is also extended heterogeneity non-temporal across units. We tackle second shortcoming by deriving ensures conditionally conjugate posterior updates all parameters. this end, take advantages Pólya-Gamma augmentation forward filtering backward sampling algorithm. After validating properties Monte Carlo study, proposed specification are demonstrated an empirical application uncover relationships between spanning nine years pavement characteristics. Model exhibit practically significant (i.e., instability). For example, safety benefits better ride quality estimated increase over time.