Bibliography

(eng): The paper addresses the task of modeling and predicting count data, with the application to traffic counts at selected urban roads in Prague. The methodology proposed in the paper presents the following key ideas: (i) analysis of explanatory multiple counts and detection of their locations through recursive Bayesian estimation of Poisson mixtures, (ii) estimation of the target count model via local Poisson regressions at recognized locations, and (iii) prediction of target counts through real-time location detection. The algorithm's properties are first investigated using simulated data and then validated with real traffic counts. Experimental results demonstrate that the proposed algorithm outperforms alternative methods in predicting traffic count data across various quality metrics, even for weakly correlated data. The main contribution of the paper is the development of a novel approach for online target count prediction, which simultaneously analyzes the spatial locations and temporal evolution of multivariate explanatory count data.