Análisis de Accidentes de Tránsito

The problem

Road accident records in Acacías, Meta were stored as raw CSV files with inconsistent date formats, duplicate neighborhood names due to capitalization and accents, and latin-1 encoding issues. There was no way to quickly answer questions like which neighborhoods had the highest accident rates, how mortality evolved over time, or what types of accidents predominated — all critical for local road safety policy.

The solution

We built a data cleaning pipeline in Python that normalizes dates, deduplicates neighborhoods and fixes encoding, producing a clean dataset ready for analysis. On top of that, a Streamlit dashboard lets users filter by year and neighborhood to see real-time KPIs (accidents, deaths, injuries, mortality rate), trend charts, accident type breakdown, a top-15 critical neighborhoods chart, and an expandable detail table. A PDF export with key indicators is available in one click.