Policy briefs
Big data y machine learning
Todas · Econometría · Big Data · Pobreza · Macro/Sectorial · Diseño de Políticas · Experimentos
-
Doblete sísmico (y parte 4): Resultados de la encuesta de verificación en terreno de la magnitud y extensión del daño de las edificaciones afectadas por el terremoto del 24J
A field survey of 1,000 randomly selected buildings in seven coastal parishes of La Guaira, carried out in August 2026 with the EMS-98 damage scale, finds that 57.4% of the 15,339 eligible buildings suffered some damage, 22.6% (about 3,466) severe damage and 11.2% (about 1,720)…
-
Doblete sísmico (parte 2): población vulnerable, distribución de daños y priorización territorial en La Guaira
Social vulnerability and earthquake damage do not spatially coincide — 112,225 vulnerable people live in undamaged buildings, meaning a response calibrated on physical damage alone would miss them. A Social Vulnerability Index projected to census-segment level via machine-learning small-area estimation finds 187,305 people (60.3% of…
-
Doblete sísmico: población afectada y costos de reposición de la infraestructura habitacional de la línea costera de La Guaira
Of an estimated 403,784 people across seven coastal parishes, 73,524 lived in damaged structures and 37,611 in severely compromised or destroyed ones
-
Violencia basada en género en la política venezolana: Un análisis de datos en redes sociales
Beyond receiving 41% more violent tweets than male counterparts, women political activists respond differently: harassment appears to inhibit their subsequent posting while not inhibiting men’s, effectively truncating their political speech. Women were 43% more likely to receive a violent interaction, and sexualising, instrumentalising and direct…
-
¿Cuántos habitantes tiene Venezuela?: Una estimación a partir de imágenes satelitales y luces nocturnas.
Venezuela’s 2020 population is estimated at 28,142,489 (±3.9%), some 14% below the official projection of 32.6 million. With no census since 2011 and roughly six million emigrants, the study trains a predictive model on census-era high-resolution imagery and applies it to 2020 satellite imagery across…