USAID-DHS Post-Neonatal Mortality Predictors Project

This is the code repository for the USAID + DHS "MCH ML" project for predictors of child mortality at 1-59 months of age.

Code structure

• config.yaml: The project configuration file, to be used with the versioning R package.
• r-package/: This contains project functions structured as an R package. You can load all package functions in your R session using devtools::load_all("/path/to/usaid-mch-ml/r-package/"), which will give access to the functions and their documentation (via help(function_name))
• data-prep/: Scripts used to prepare the data used for analysis. Saved to the "prepared_data" folder in the config.
• run-model/: Scripts used to run the machine learning models and visualize model results. Saved to the "model_results" folder in the config.
• exploratory/: Scripts used for one-off testing and simple use cases.

Useful links

Data access:

• Project SharePoint folder (request access from Rachael Church)
• DHS data platform
• ACLED data platform

Methods:

• caret package documentation and available models
• SAGE introductory paper and blog post
• SAGE Python repository