QGIS pypopRF Plugin

A QGIS plugin for high-resolution population mapping using machine learning and dasymetric techniques. Create detailed population distribution maps by combining census data with geospatial covariates.

Features

• Create high-resolution population maps from census data
• Use building data and other geospatial information as predictive variables
• Automated machine learning workflow with Random Forest
• User-friendly interface integrated into QGIS
• Parallel processing support for large datasets
• Real-time progress monitoring and logging
• Support for age-sex population structure mapping

Requirements

• QGIS 3.0 or later
• Python 3.9 - 3.12
• Plugin dependencies will be installed automatically during installation

Installation

1. Download the plugin ZIP file from the GitHub repository
2. In QGIS, go to "Plugins" → "Manage and Install Plugins" → "Install from ZIP"
3. Select the downloaded ZIP file
4. During installation, a console window will open showing the automatic installation of required Python packages. Please do not interrupt this process as it may take several minutes.

Note: Installation through the official QGIS Plugin Repository will be available soon.

Input Data Requirements

Required Files:

• Mastergrid: GeoTIFF file defining analysis zones
• Census Data: CSV file with population counts
• Covariates: One or more GeoTIFF files (e.g., building data)

Optional Files:

• Age-Sex Data: CSV file with age-sex population structure
• Water Mask: For excluding water bodies
• Constraints: Additional spatial constraints

Outputs

The plugin generates the following files in your project's output directory:

Population Distribution:

• normalized_census.tif: Normalized census values
• population_unconstrained.tif: Default population distribution output
• population_constrained.tif: Distribution with constraints (when provided)

Machine Learning:

• model.pkl.gz: Trained Random Forest model
• scaler.pkl.gz: Feature scaler
• features.csv: Extracted features with importance metrics

Additional Outputs:

• agesex/: Age-sex structure outputs (when age-sex data provided)
• Detailed processing logs

Getting Help

• Documentation: https://wpgp.github.io/QGIS-pypopRF/
• Issues & Support: https://github.com/wpgp/QGIS-pypopRF/issues
• WorldPop SDI: https://sdi.worldpop.org

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

If you use QGIS-pypopRF in your research, please cite:

@software{QGIS-pypopRF,
  author = {Nosatiuk B., Priyatikanto R., Zhang W., McKeen T., Vataga E., Tejedor-Garavito N, Bondarenko M.},
  title = {QGIS-pypopRF: Population Prediction and Dasymetric Mapping Tool},
  publisher = {GitHub},
  url = {https://github.com/wpgp/QGIS-pypopRF}
}

Development Team

Developed by the WorldPop SDI Team:

• Borys Nosatiuk (b.nosatiuk@soton.ac.uk) - Project Lead
• Rhorom Priyatikanto (rhorom.priyatikanto@soton.ac.uk)
• Maksym Bondarenko (m.bondarenko@soton.ac.uk)
• Wenbin Zhang (wb.zhang@soton.ac.uk)
• Tom McKeen (t.l.mckeen@soton.ac.uk)
• Elena Vataga (e.vataga@soton.ac.uk)
• Natalia Tejedor Garavito (n.tejedor-garavito@soton.ac.uk)