ROBUST MINIMUM COVARIANCE DETERMINANT SCALE FOR ADDRESSING OUTLIERS IN FOOD SECURITY INDEX DATA

Abstract

Food Security Index (FSI) data can present challenges in the analysis due to data diversity, such as outliers. This study aims to address this issue by employing robust regression with the Minimum Covariance Determinant Scale (MCD-S) estimator on the 2023 FSI data from Papua, Indonesia. The research used secondary data from the 2023 Food Security and Vulnerability Atlas report, with a sample of 42 regencies and cities in Papua. From the result, it found that seven independent variables significantly impact the FSI in Papua for 2023: are the ratio of normative consumption per capita to food availability (X₁), the percentage of people living below the poverty line (X₂), the percentage of households with a proportion of expenditure on food of more than 65 percent (X₃), the percentage of households without access to electricity (X₄), the average length of schooling for women over 15 years (X₅), the percentage of stunted toddlers (X₈) and life expectancy (X₉). The coefficient of determination value of 0.9998 suggests that 99.98% of the variability in strength can be attributed to the regression model. This study contributes to the limited research on robust MCD estimators, particularly MCD-S, in addressing outliers in FSI data.