Visual analytics for public health: exploring interpersonal violence and depressed mood among adolescents

This study introduces a replicable framework integrating EDA, statistical inference, and visualization to support data-driven decisions on Brazilian adolescents’ mental health. Using the 2019 (PENSE) survey microdata, it examines associations between sociodemographic factors, interpersonal violence, and depressed mood. Analyses include statistical tests, combined with multilevel visualizations. Results show territorial inequalities and significant differences by sex, age, and race. Recurring associations emerge with coercion, abuse, physical aggression, bullying, and cyberbullying. Key challenges include class imbalance and ambiguity in responses categorized as "other motives". Overall, EDA and visualization help prioritize attributes, identify potential biases, and support a reusable public-health pipeline for similar studies.