Revealing the hidden pattern of under-five malnutrition prevalence distribution in West Java-Indonesia from canonical correspondence analysis and predictive clustering perspective

Abstract

The global burden of under-five malnutrition has become a critical and urgent health and welfare problem to the present. Since under-five malnutrition has varying determinant factors, this study offers another perspective to reveal the hidden patterns of the malnutrition prevalence distribution among under-fives by exploring their associations. Canonical correspondence analysis and predictive clustering were employed to (1) reveal the association between the under-five malnutrition prevalence with nutritional intake, environmental health, spatial, and socio-economic factors and (2) predict the clustering of under-five malnutrition emergency status for a certain region. The results suggest that (1) there exists a statistically significant association between the prevalence of under-five malnutrition and its determinant factors, and (2) three clusters have been identified to represent the under-five malnutrition emergency status (critical, serious, warning) and a model has been constructed with an accuracy rate of 83.3% in predicting a certain region status with given attribute variable values. The designation of under-five malnutrition emergency status is desired to provide direction regarding the priority level in handling malnutrition for certain regions.