• Amali V. E.
  Joseph Sarwuam Tarka University, Makurdi, Benue State, Department Physical Sciences
  




• Nwaosu. S.C.
  Joseph Sarwuam Tarka University, Makurdi, Benue State, Department Physical Sciences
  




• Kuhe A.D
  Joseph Sarwuam Tarka University, Makurdi, Benue State, Department Physical Sciences
  




• Udomouh . E. F.
  Joseph Sarwuam Tarka University, Makurdi, Benue State, Department Physical Sciences
  

This study applied a Bayesian Vector Zero-Inflated Generalized rnPoisson INGARCH (ZIGP-INGARCH) model to captures the rnmultivariate dynamics of COVID-19 time series data in rnNigeria, characterized by overdispersion, excess zeros, and rninter-variable dependence. The analysis utilizes daily counts of rnnew cases, deaths, and recoveries from March 2020 to May rn2023, obtained from the Nigeria Centre for Disease Control rn(NCDC). The model jointly estimated the conditional means of rnthe three epidemiological indicators while accounting for both rntemporal and cross-sectional dependencies using a flexible rnINGARCH framework embedded with zero-inflated rngeneralized Poisson marginals. Bayesian inference is rnperformed using Hamiltonian Monte Carlo (HMC), ensuring rnrobust parameter estimation and uncertainty quantification. rnResults reveal that the ZIGP-INGARCH(1,1) model captures rnsignificant moving average dynamics in recovery-to-case rninteractions ( B₁₃ = 0.44), mild overdispersion in new and rnrecovered cases (ϕ₁ ≈ 0.81, ϕ₃ ≈ 0.80), and a moderate zerorninflation probability for death counts (π₂ ≈ 0.45). Model rnadequacy was confirmed through residual diagnostics, rnAIC/BIC comparisons, and convergence metrics (R-hat ≈ rn1.000). This research demonstrates the utility of Bayesian rnZIGP-based multivariate models in understanding pandemic rndynamics and guiding effective forecasting and policy rnresponse.