Predicting subsequent childbirth plans of Korean mothers using machine learning: insights from the Panel Study on Korean Children

Global fertility rates have reached critically low levels, at 1.58 in 2022 OECD countries, while South Korea’s rate plummeted to 0.72 in 2023. The machine learning models developed in this study predict subsequent childbirth intentions among Korean mothers, thereby diverging from previous research that focused on traditional analyses of individual variables. The models demonstrated robust performance in predicting mothers who will or will not have more children, but they showed limited efficacy for the “undecided” group. Among the numerous policy interventions, most have proven insufficient in reversing South Korea’s fertility decline. We propose that these predictive models would aid in tailoring governmental policies that would equally support women who decide to have more children and those who choose not to, as these groups have distinct needs. This targeted approach may offer a more nuanced strategy for addressing low fertility rates in South Korea and potentially other nations facing similar demographic challenges.