Prediction of gender power dynamics and political representation in Nigeria using machine learning models

Abstract

This study applies machine learning to investigate gendered power dynamics and women's socio-economic and political engagement in Nigeria (1991-2023, World Bank, UNDP, INEC data). We trained Random Forest, Support Vector Machine (SVM), and Neural Network models with k-fold cross-validation, evaluating performance with R², RMSE, and MSE. The SVM model demonstrated superior performance (R² = 0.998). Feature analysis revealed that women's industry participation positively correlates with population share and education, while rural residence diminishes their likelihood of being employers. Additionally, K-means clustering of 2023 voting data uncovered regional variations in women's political representation. This research highlights enduring socio-economic and spatial barriers, demonstrating how AI-based evidence can inform gender-sensitive policies for inclusive representation in Nigeria.