Análise temporal e causal das chamadas ao disque 180 e notícias do Ministério da Mulher: um estudo com técnicas de machine learning

Abstract

This study explores the influence of news from the Ministry of Women on the number of calls to the Disque 180 helpline, utilizing a quantitative approach that combines time series techniques with advanced statistical models. Time series of news, both with and without a specific filter on the theme of violence against women, were analyzed to assess the direct impact on public re porting behavior. Methods included gaussian process, Granger causality, vector autoregressive (VAR) model and the transfer entropy. The application of a thematic filter to news data resulted in more accurate forecasts and a stronger identified relationship between variables, highlighting that focused campaigns and information directly related to violence against women elicit a more immediate and consistent response in calls to Disque 180. These results underscore the impor tance of segmented and specific public communication to positively influence social reporting behavior, providing valuable insights for awareness and anti-violence policy efforts.