Dynamics and Temporal Persistence of Femicide in Brazil Under Conditions of Uncertainty (2017–2025)

This article examines the temporal dynamics of femicide in Brazil between 2017 and 2025 through the identification of relatively stable temporal regimes and the probabilistic modeling of count data under conditions of uncertainty. The series exhibits non-linear behavior characterized by initial expansion, relative deceleration, and moderate resumption of growth in recent years. To capture these dynamics, the study employs a Generalized Additive Model with Negative Binomial distribution (GAM-BN), an approach particularly suitable for modeling flexible temporal variation in count data. Rather than relying on deterministic interpretations, the proposed framework analyzes the evolution of the series through probabilistic trajectories conditioned by the observed temporal dynamics. The findings suggest the presence of distinct temporal regimes in the evolution of femicide, including accelerated initial growth, partial stabilization at elevated levels, and moderate reintensification in recent years. Conditional projections for 2026 organize probabilistic scenarios associated with the continuity of the observed dynamics. Substantively, the results remain consistent with the structural persistence of gender violence at persistently elevated levels despite gradual variations in growth intensity. The article contributes by integrating temporal interpretation, probabilistic modeling, and sociological analysis into an analytical framework aimed at understanding the continuity of violence under conditions of uncertainty.