<![CDATA[Food security modeling in Indonesia needs to consider interregional linkages and strong correlations among predictor variables. However, the application of spatial models that simultaneously address spatial dependence and multicollinearity at the provincial level remains limited. This study aimed to apply the Spatial Autoregressive Model (SAR) to model Indonesia’s 2024 Food Security Index (FSI), using Principal Component Analysis (PCA) as an approach to address multicollinearity. The study employed an applied quantitative approach using secondary data from 38 provinces obtained from Badan Pangan Nasional and Badan Pusat Statistik. Seven predictor variables were analyzed using multiple linear regression, variance inflation factor (VIF), PCA, Moran’s I, Lagrange Multiplier, and SAR with a k-nearest neighbors (KNN) spatial weight matrix. The results indicated high multicollinearity among three predictor variables. PCA produced three principal components that explained 94.50% of the variance in the data. A Moran’s I value of 0.71005767 indicated positive spatial autocorrelation, while the SAR spatial coefficient of 0.52676 indicated a significant spatial effect. The SAR model performed better than multiple linear regression, with an AIC value of 242.5907 and an R² of 0.7577. These findings confirm that integrating PCA and SAR produces FSI modeling that is more consistent with the characteristics of the data and can support the formulation of food security policies that consider interprovincial linkages. Future research may develop a spatial panel approach to analyze FSI dynamics over time.]]>
Copyrights © 2026
