Modeling the Percentage of Poor Population in West Sumatra in 2024 Using Nonparametric B-Spline Regression

Poverty is a multidimensional issue and remains a major challenge in West Sumatra Province. This study analyzes the percentage of the poor population in 2024 using a nonparametric B-Spline regression approach with the independent variables being the percentage of per capita expenditure on food, the labor force participation rate, and the average length of schooling. Data are sourced from the official publication of the Central Statistics Agency of West Sumatra Province in 2024. The results show that the best model is obtained from the second-order B-Spline regression with one node for each independent variable, based on the minimum Generalized Cross Validation (GCV) value. This model produces a coefficient of determination of 92.46% and a Mean Absolute Percentage Error (MAPE) of 11.02%, indicating high prediction accuracy. Substantively, the average length of schooling has a negative effect on the poverty rate, while food expenditure and labor force participation have varying effects across regions. These findings indicate that B-Spline regression is effective in capturing nonlinear and complex relationships between socioeconomic variables, but its interpretation still needs to consider the empirical context in the field.