DNP-XGboostGarzon

XGBoost classification model outputs estimating building footprints from satellite imagery for the Municipality of Garzón. The dataset was developed during the DataSandbox Colombia 2020 by the Data Science Unit of the National Planning Department. It includes probability scores and polygon geometries for each detected construction.

Use Cases

• Validate building detection accuracy by comparing probability scores against ground truth data.
• Analyze urban density and settlement patterns using the polygon geometries for Garzón.
• Train or fine-tune other machine learning models for land cover classification using the probability and polygon columns.
• Create thematic maps of estimated construction coverage for the municipio of Garzón.

Strengths

• Includes a probability column for each prediction, allowing for confidence-based filtering.
• Data is projected in MagnaSirgas 3115, a standard coordinate reference system for Colombia.
• Available in multiple open formats including CSV, JSON, XML, and RDF.

Limitations

• Sample data is unavailable, making initial quality assessment difficult before download.
• Key metadata such as row count, license, and author are unknown.
• Freshness should be verified; the last update timestamp is 2026-05-18 18:45:02.

Provenance

Source

Data Science Unit of the National Planning Department (Colombia).

Collection Method

Derived from applying an XGBoost classification model to satellite imagery.

Freshness

Last updated 2026-05-18 18:45:02.

Geography

Municipality of Garzón, Colombia.