APPLICATION OF MULTIVARIATE ADAPTIVE REGRESSION SPLINES (MARS) TO MODEL THE FACTORS AFFECTING THE PERCENTAGE OF POOR POPULATION IN INDONESIA

Nur Shanty (Universitas Negeri Makassar)
Ruliana (Universitas Negeri Makassar)
Muhammad Kasim Aidid (Universitas Negeri Makassar)

VARIANSI: Journal of Statistics and Its Application on Teaching and Research
Vol. 7 No. 03 (2025)

Poverty is one of the social and economic problems that Indonesia continues to face today. The Multivariate Adaptive Regression Spline (MARS) is a nonparametric regression model that estimates the functional relationship between the response variable and predictor variables when the relationship form is unknown. This study aims to estimate the parameters of the Multivariate Adaptive Regression Spline (MARS) method for the percentage of poor population in Indonesia and to identify the factors that significantly affect the percentage of poor population. The results of this study found that the best model was obtained with a combination of BF = 21, MI = 1, and MO = 3, with GCV = 0,3102717. Based on the MARS model, the variables that significantly affect the percentage of the poor population are the percentage of formal workers (x3), percentage of households with access to proper sanitation (x4), and Gini Ratio (x7) with a coefficient of determination (????²) of 81,44%.