Pemodelan Prevalensi Stunting di Indonesia Menggunakan Metode Regresi Nonparametrik Spline Truncated (Studi Kasus: Data BPS Tahun 2023)

Abstract

Regression analysis is a statistical method used to determine the pattern of relationship between one variable and another. When the shape of the regression curve is unknown, nonparametric regression is used. One approach to nonparametric regression is truncated spline nonparametric regression. Truncated splines work by dividing the data into several intervals and limiting them with knot points. This allows the truncated spline to handle changes in data patterns and create a model that tends to be flexible. The case study in this research is the prevalence of stunting in Indonesia. The decline in stunting cases in Indonesia tends to be slow and still exceeds the maximum standard limit of 20%. Therefore, this study aims to model the prevalence of stunting in Indonesia and identify the factors that significantly influence it using 2023 BPS data. Through the seven variables used, it was found that pregnant women who received supplementary food programs, pregnant women who underwent antenatal care (K4), infants born within 60 minutes who received early breastfeeding initiation, infants aged 0–23 years who received exclusive breastfeeding, infants aged 0–11 months who received complete basic immunization, children aged 6–23 months who consumed a minimum variety of foods, and households with access to adequate sanitation significantly influence stunting prevalence. The model's coefficient of determination is 98.83%, and the lowest GCV at three knot points is 8.16.