Machine Learning-Based Fertility Decline Analysis and Forecasting in India

This paper employs a machine learning-based framework to analyze and forecast fertility decline in India. Using secondary data from NFHS surveys, the Census of India, and global demographic databases (1950–2021), it examines long-term fertility trends, regional disparities, and socio-economic determinants, with a focused case study of Andhra Pradesh. Fertility indicators such as Total Fertility Rate (TFR), General Fertility Rate (GFR), and Crude Birth Rate (CBR) are analyzed alongside variables like female education, urbanization, and contraceptive use. India’s TFR has declined from 6.2 in 1950 to 2.0 in 2021, with Andhra Pradesh recording a lower 1.5 and a 17% drop in GFR over the past decade. Predictive modeling using Linear Regression, Random Forest, SVM, and XGBoost reveals that XGBoost provides the most accurate forecasts, projecting a continued decline through 2024. While this transition reflects social and health progress, it also signals challenges of population aging and dependency. The study underscores the utility of machine learning in demographic forecasting and evidence-based policy making.Unlike previous research that focused on clinical fertility outcomes or narrow area forecasts, this work takes a national-level, long-term perspective (1950–2021), incorporates socioeconomic and environmental variables, and confirms XGBoost's better performance in demographic prediction.