Analyse exploratoire du dataset World Bank Identification for Development (ID4D). Cette présentation couvre la distribution mondiale de la possession d'identité légale, les fractures d'accès (genre, urbain-rural, revenus), l'évolution temporelle, l'identité numérique et un clustering K-means des profils de pays. Réalisé dans le cadre du cours d'Analyse Exploratoire de Données (EDA) sous la supervision de Mr AZEYOU.

Repeated-measures ANOVA is among the most widely used designs in clinical and behavioral research — tracking the same subjects over time or across conditions. Yet one of its central assumptions, sphericity, is routinely misunderstood, mechanically tested with an inadequate test, and patched with a correction that does not fix the underlying problem.

Nematodirus infection intensity in Wyoming sheep is a useful applied setting for comparing categorical and count-based modeling strategies. The response is the number of sheep in ordered infection-intensity categories within each farm. A multinomial model is a natural starting point because the counts are allocated across mutually exclusive categories, but a log-linear formulation is more flexible for modeling overdispersion, farm-level clustering, and zero inflation. This report presents a staged analysis using multinomial, Poisson, negative binomial, heterogeneous-dispersion negative binomial, mixed-effects negative binomial, and zero-inflated negative binomial models. The final selected model was a negative binomial log-linear model with heterogeneous dispersion and a farm-level random intercept.

Danna, Jhonny, Jacobo