Implementation of K-Nearest Neighbors Algorithm on Regional Food Security Classification in Indonesia

Abstract

Classifying regions that are vulnerable to food security in the face of a global recession is critical. Thus, this research aims to classify regions in Indonesia based on food security characteristics. The sample used is 34 provinces grouped by geographical area in Indonesia. The food security variables used include food availability, access, and absorption. The K-nearest neighbors (KNN) classification was applied to the data using two nearest neighbors. All provinces in the region of Java and Sulawesi are predicted to have high food security.
In contrast, several provinces in Sumatera, Kalimantan, Nusa Tenggara, Maluku, and Papua still have relatively low food security. Papua has the lowest food security rate for availability, access, and utilization. In conclusion, the KNN model in this study has a fairly good performance with an accuracy rate of 76.47%. This is obtained by considering the amount of data, the size of the training data, and the model complexity of the many features.