changemaker

ResearchInnovations for Poverty Action25 min read

Replication package for: Predicting Poverty-Resilient Households using Machine Learning in Sub-Saharan Africa

Nouréini Souleymane Sayouti, Edouard Mensah

In short

This research provides a complete replication package for predicting poverty-resilient households in Nigeria using machine learning models. Combining World Bank panel data with climate and conflict metrics, the study reproduces all published figures and tables.

Three takeaways

  1. 01

    Run python reproduce.py to automatically rebuild all analysis files and figures from the original data.

  2. 02

    Access shareable datasets containing conflict measures, climate values, deflators, and sample identifiers without redistributing restricted files.

  3. 03

    Rely on pinned dependencies and checksums to refit models matching reference outputs to extreme precision.

Why it matters for your work

Incorporate these reproducible workflows into your poverty targeting projects to verify machine learning models and ensure your vulnerability analyses withstand independent scrutiny.

Read the original (free) at Innovations for Poverty Action

Add it to your portfolio

One line in your own words. It shows on your page under “Always learning”.

Keep what you learned.

Write one takeaway and it shows on your portfolio page, dated, linked to the source.

Sign in to add it to your page