This repository houses a collection of scripts designed for top-down population disaggregation using the randomForest R package and the **bartMachine R package **. It's aimed at disaggregating population count from a higher administrative level to small area levels.
There are four main scripts included in this repository:
- BART_Simulation_Workflow.R: This script implements a simulation study using Bayesian random forest disaggregation. It explores the capabilities of the Bayesian approach in population disaggregation.
- RF_Simulation_Workflow.R: This script conducts a simulation study using the randomForest R package. The results from this simulation study are compared to the Bayesian random forest approach to assess the predictive performance of both models.
- BART_Workflow.R: This workflow applies the Bayesian approach to disaggregate the 2021 census data of Ghana. It demonstrates the practical application of Bayesian random forest in population disaggregation.
- Random_Forest_Workflow.R: This workflow employs the randomForest package to disaggregate the 2021 census data of Ghana. It showcases the application of the random forest modelling approach in the same context.
To get started with the scripts and workflows in this repository:
- Clone this repository to your local machine.
- Install the necessary R packages, including randomForest and bartMachine, if you haven't already.
- Run the desired script or workflow to explore top-down population disaggregation using the specified modelling approach.
Contributions and enhancements to this repository are welcome. Feel free to submit issues, fork the repository, or create pull requests to collaborate and improve the scripts and workflows.