Reproducible
Scripts and CLIs you can version, cite, and rerun when datasets or methods evolve.
Open research · Shared code
Open source projects
Three kofiyatech repositories you can fork, study, and extend—spanning data engineering, applied machine learning, and reproducible research tooling you can share with the community.
Built in the open
These projects ship as source code so students, researchers, and partners can audit methods, rerun pipelines, and contribute improvements—without waiting on a black box.
- Prepare data
- Model & analyse
- Publish findings
Community-first
Issues and pull requests welcome—graduate cohorts and NGOs use the same entry points.
Research-ready
Built for real studies and teaching—not demos. Several tie into our DHS AI/ML Toolkit where survey data is the focus.
Repositories
Explore the trio
Pick a project below—read the launch story in our Newsletter or jump straight to GitHub to contribute.
-
01 Python Data engineering · Open sourceDHS-To-Database-dhs2CSVTables-simplified
A friendly wrapper around Harry Gibson’s DHS-To-Database that converts raw DHS drops into relational CSV tables—ready for SQLite, PostgreSQL, and downstream ML pipelines.
- Python 3.8+
- CSV → SQLite
- csvs-to-sqlite
-
02 Python Machine learning · Child survivalMintilo AI
Predictive modelling for under-five survival using household, spatial, and survey metadata—with tools to explore feature relevance and compare countries.
- Scikit-learn
- DHS surveys
- Policy insights
-
03 R Survival analysis · Open sourceWatoto Survival
Classical survival methods—Kaplan–Meier and Cox regression—for collaborating researchers studying under-five mortality risk factors across DHS countries.
- Kaplan–Meier
- Cox regression
- R · DHS
Working with Demographic and Health Surveys? The DHS AI/ML Toolkit maps the wider research stack—dashboards, deep learning, and Bayesian tooling alongside these repos.
Open toolkit