Environmental risk assessment often requires modeling complex temporal processes influenced by multiple variables and characterized by environmental condition shifts. This work introduces a novel methodological framework based on concomitant-variable multivariate penalized hidden semi-Markov models (CV-MPHSMM) with autoregression to capture such dynamics. The proposed model extends traditional hidden semi-Markov models by integrating concomitant variables to account for external environmental factors influencing state transitions and sojourns, and by incorporating penalization techniques to enhance model interpretability and prevent overfitting in high-dimensional settings. Autoregressive components are included to model temporal dependencies within and between observed multivariate time series. Analytical expressions for multivariate risk measures are obtained under the CV-MPHSMM. The framework is applied to pollution, demonstrating its capacity to identify latent states, quantify transition probabilities, and detect environmental condition shifts. Simulation studies validate the robustness and flexibility of the proposed model in handling complex scenarios, while case studies highlight its practical utility in informing risk management strategies. The findings underscore the potential of CV-MPHSMMs with autoregression as a powerful tool for advancing environmental risk assessment and decision-making under uncertainty.
7 marzo 2025, ore 14.00
Antonello Maruotti
Università LUMSA
In person: Room 34 (4th floor) building CU002 Scienze Statistiche
Link Zoom
https://uniroma1.zoom.us/j/83625004899?pwd=bXCtz0mp759PUh2lkqT0BUoVa0Uegg.1
ID riunione: 836 2500 4899
Passcode: 123456