ToniaNiles/maternal-risk-prediction-refactored

Refactored MSc dissertation project exploring maternal health vulnerability in Nigeria using machine learning, geospatial analysis, and reproducible environments. # Predicting Maternal Mortality Risk in Nigeria with Machine Learning and Geospatial Analysis **MSc Dissertation Refactoring Project - Risk Classification, Geospatial Analysis and Exploratory Policy Scenarios** This repository contains the analytical workflow developed for my MSc dissertation, which examines maternal health vulnerability in Nigeria using machine learning, socioeconomic and healthcare access indicators, geospatial analysis, and exploratory policy scenarios. The project moves beyond a clinical-only perspective on maternal health to investigate how broader socioeconomic and systemic factors—including education, healthcare access, health insurance coverage, mobile phone ownership, antenatal care, and place of delivery relate to maternal vulnerability. --- ## Project Overview Maternal mortality remains a major public-health challenge in Nigeria. While direct medical causes such as haemorrhage and eclampsia are well documented, maternal outcomes are also shaped by wider socioeconomic, demographic, and healthcare-access conditions. This project applies a data-driven approach to examine these upstream factors using nationally representative survey data from Nigeria. The analytical workflow combines descriptive analysis, supervised machine learning, class-imbalance handling, model optimisation, feature-importance analysis, geospatial analysis, and exploratory scenario simulations. Three supervised classification algorithms were evaluated: - **Logistic Regression** - **Random Forest** - **XGBoost** Following model comparison and optimisation, the tuned Random Forest demonstrated the strongest overall predictive performance and was used for subsequent model interpretation and exploratory scenario analysis. --- ## Dataset The analysis uses data from the **Nigeria Multiple Indicator Cluster Survey (MICS)** provided by UNICEF. The original survey data were cleaned and prepared to retain variables relevant to the study's focus on maternal vulnerabili …