Recognizing Patterns of Food Insecurity in Indonesia New Recommendations for Hunger Prevention Policy

Abstract

Geographically, Indonesia is a tropical area that has two seasons so it is very suitable for agriculture. Indonesia is also known as an agricultural country. It means that most of its population works in the agricultural sector which is related to food. However, food insecurity still occurs in Indonesia. This is supported by the COVID-19 pandemic in which the Food and Agriculture Organization (FAO) stated that there is a threat of food scarcity during the COVID19 pandemic. This will interfere with the SDGs second goal of ending hunger and sustainable agriculture. The purpose of this study captures the causes of food insecurity in each province in Indonesia spatially using the bicluster method. The data that used is data from Susenas and Sakernas by BPS in 2019. The results of three studies using the bicluster method show that each province has different aspects and letters. The results of this research can serve as a reference for policies in dealing with hunger in Indonesia. This study uses the CC Algorithm method by Cheng and Church. In the bicluster approach, we run some parameters and do manual tunning to choose the best parameter based on the average of residue over volume (MSR/V). CC algorithm tries to get bicluster as large as possible with low MSR Value, therefore the best parameter should be the one that generates the smallest MSR/V and the smallest MSR/V is produced when sigma = 0.3. The application of the CC biclustering algorithm to the poverty structure in Indonesia produces 3 biclusters. Bicluster 1 Consist of 28 Province with 11 Variables, Bicluster 2 consists of 4 Province with 7 Variables, and Bicluster 3 consist of 2 Province with 12 variables. Variable X5 and X6, as absolute food insecurity measurement is included in all bicluster.