Univariate model to describe and forecast the behavior of the historical series of femicides in Peru

Abstract

This research, with an analytical-applied approach, was developed in the region of Puno, Peru, with the purpose of projecting femicide in the country. Femicide, a global concern, places Peru in the bottom third of Latin America in terms of rate. The importance of the study lies in anticipating this phenomenon in order to implement effective preventive strategies. The main objective was to identify a monthly predictive model that fits the historical series, optimizing the allocation of human resources, equipment and technology for better decision-making in the short term. The population covered femicides recorded since 2001, using data from the AURORA National Program between 2010 and 2021 as a sample. Using probabilistic time series analysis, the most appropriate univariate model to predict the behavior of femicide was identified: Y_t= 10.16817 + 0.15179ε_(t-1) + 0.54948ε_(t-12) + 0.77771ε_(t-12) + ε_t . This model, based on the Box-Jenkins methodology, met all statistical requirements, providing a reliable fit for the monthly historical series. The results obtained are key to designing public policies aimed at preventing and reducing femicide in the country.