Final Projects
Projects Catalog
Spatial and multi-temporal analytics system for early warning detection of labour deterioration in Colombia
This project develops an early warning system based on spatial and multi-temporal analytics to map the deterioration of labour conditions and structural informality in Colombia. To optimise the targeting of public policies, a Big Data architecture (a lakehouse ecosystem on Polars) was implemented to process microdata from the Great Integrated Household Survey (GEIH 2022–2025 by DANE) and cross-reference it with infrastructure cartography from OpenStreetMap (OSM). The methodology evaluates the inter-annual variation of underemployment and the risk of economic inactivity, ensuring rigorous statistical secrecy through differential privacy (Laplace noise injection). The results demonstrate that labour deterioration manifests primarily as a chronic transition towards informality, presenting a strong spatial correlation with infrastructure deserts (lack of connectivity and services). The system provides an interactive dashboard that functions as a budgetary triage tool, recommending that the government prioritise comprehensive interventions exclusively in departmental clusters where high demographic vulnerability and structural abandonment converge.
Youth labour risk in Colombia: territorial, temporal, and gender analysis, GEIH 2022–2025
This project analyses youth labour risk in Colombia for people aged 18 to 28 during 2022–2025 using monthly microdata from DANE’s Great Integrated Household Survey (GEIH). The pipeline reconstructs raw, processed, and curated layers; harmonises demographic, labour-force, and employed-person modules; validates observed codes; and applies expansion factors to estimate population-level indicators. The core analytical products are monthly unemployment, informality, and employment rates by department, gender, and urban–rural area. A composite youth labour risk index combines normalised unemployment and informality with a 60/40 weighting scheme, allowing territorial, temporal, and gender comparisons. The project also identifies departments with higher average risk and larger gender gaps, builds a Colombia choropleth map, and exports a static HTML dashboard organised around the research objectives. To support explainability, a descriptive linear regression model summarises partial associations between observed labour risk and month, department, zone, gender, age, and temporal trend. The model is not interpreted causally. Governance controls include a manifest, audit log, schema contract, reproducible download/cache logic, and k-anonymity suppression with k=30 for public tables. Overall, the project turns GEIH microdata into governed, reproducible evidence for prioritising youth employment policy by territory, gender, and zone.