African Journal of Biomedical Research

Résumé

This study uses the Hamiltonian Monte Carlo (HMC) and Gibbs methods to predict of Hypertension, Tobacco use, Overweight and Obesity. The methods were applied on data from World Health Organization report 2015. Corresponding stochastic matrices were generated and utilized in the prediction and comparison of the two sampling methods. Using Python and the NumPy library, the state vectors for these conditions were predicted, with the results showing slight variations between the two methods. In the long term, both methods converged to stable probability distributions, indicating steady-state values for each condition. HMC predicted long-term prevalence of 17.48% for hypertension, 17.5% for tobacco use, 17.56% of Overweight, 14.36% of Obesity and 33.1% for "Other subjects, while Gibbs predicted 17.82%, 26.27%, 17.13%, 4.80% and 33.98%, respectively. Additionally, Gibbs was found to outperform HMC in terms of error metrics, with lower values for Mean Squared Error (MSE), Mean Absolute Error (MAE), indicating better prediction accuracy. This analysis aids practitioners in selecting the most suitable method based on specific problem characteristics and dataset complexities.

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0