Skripsi
PEMODELAN REGRESI SPASIAL PREVALENSI KETIDAKCUKUPAN KONSUMSI PANGAN BERDASARKAN KABUPATEN/KOTA PADA PULAU DI INDONESIA
This study aims to model and analyze the factors that influence the Prevalence of Undernourishment (PoU) at the regency/city level in Indonesia. The method used is spatial regression. The analysis was conducted separately for the five major islands in Indonesia, namely Sumatra, Java, Kalimantan, Sulawesi, and Papua. The data used included PoU as the dependent variable and seven independent variables, which consist of Rice Production (RP), Population Density (PD), Percentage of Poor Population (PPP), Percentage of per Capita Food Expenditure (PCFE), Desirable Dietary Pattern (DDP), Human Development Index (HDI), and Food Security Index (FSI). The results of spatial autocorrelation tests show that all five islands exhibit spatial autocorrelation, either in the dependent variable, the residuals, or both. Based on these results, the SAR model was applied to Kalimantan; the SEM model was applied to Papua; and the GSM model was applied to Java. Meanwhile, for Sumatra and Sulawesi, no single best model can be identified, as the three models show only small differences in AIC, BIC, and R² values. Therefore, the most appropriate spatial model cannot be clearly justified. The factors that significantly affect PoU are different across islands. DDP and HDI influence PoU in Sumatra and Kalimantan; PPP, DDP and HDI influence PoU in Java; PPP, DDP and FSI influence PoU in Sulawesi; and PCFE influences PoU in Papua. The spatial models indicate that PoU in a regency/city on an island is influenced by PoU in neighboring regencies/cities. These findings highlight the importance of island-specific spatial analysis, especially by accounting for neighboring regions that share direct boundaries. Keywords: Spatial Autocorrelation, PoU, Spatial Regression