MULTIVARIATE ADAPTIVE REGRESSION SPLINES MODELING FOR HOUSEHOLD FOOD SECURITY IN CENTRAL BORNEO PROVINCE 2017

Abstract

This study aims to make a model household food security with social, economic and demographic variables in Central Borneo Province using nonparametric regression, Multivariate Adaptive Regression Splines (MARS). The source of data was 2017 Socioeconomic Survey (SUSENAS) from Central Bureau of Statistics. The model shows that all variables have significant to affect household food security with importance level: area type (100%), age (99.47%), the amount of people in a household (88.57%), access to credit (81.01%), head of household education (19.74%), and gender (11.08%). In addition, there are four basic functions with no variable interaction, two basic functions with two variable interactions and two basic functions with three variable interactions on the MARS model specification.