Unveiling India's Fertility Decline Trajectory: Hybrid ML Models for Accurate Long-Term Forecasts

This research uses machine learning to anticipate fertility drop in India from 1950–2021, utilizing data from National Family Health Surveys, Census of India, and worldwide demographic datasets. The study focuses on fertility trends, regional differences, and socioeconomic variables in Andhra Pradesh. Fertility indicators such as total fertility rate, general fertility rate, and crude birth rate are studied with variables such as female education, urbanization, and contraception use. Four models (linear regression, random forest, support vector machines, and XGBoost) are trained on historical fertility data to forecast short-term until 2024. XGBoost is the most accurate forecaster, indicating a sustained drop in fertility, especially in low fertility states.The findings emphasize both the social and health benefits of decreased fertility, as well as emerging difficulties such as population aging, labor supply, and reliance. The study highlights the importance of integrating machine learning and demographic data for evidence-based policy planning in a constantly changing population context.