Modeling of poverty level in South Sulawesi Province through spline nonparametric regression approach

Abstract

This study discussed modeling poverty levels in South Sulawesi through Spline Nonparametric Regression Approach. Since no specific pattern was formed from the relation between the poverty level in South Sulawesi and the factors that influence it, the researchers used nonparametric regression modelling with Spline approach. The Spline model is very good in modelling data that has changeable patterns at certain subintervals. The aimed of this study were to investigate the factors that have the most influence on the poverty level and modelling the poverty level in South Sulawesi through nonparametric regression. The scope of this study was the use of Generalized Cross-Validation (GCV) in selecting optimal knot points; in this case, it used 1, 2, and 3 knots. This study found that three factors mainly influence the poverty level in South Sulawesi in 2017: unemployment, population growth, and literacy rates. The results showed that the best spline model was the model with three-knot points and the minimum GCV value is 11.1155. In addition, the  value is 79.75%. It means that the variables of unemployment, population growth, and literacy rate can explain 79.75% of the variation of the poverty variable, while other variables explain the remaining 20.25%.