Classification of Bird Species using K-Nearest Neighbor Algorithm

Abstract

Birds play an essential role in the functioning of the world's ecosystems by directly impacting human health, economy, and food production and benefiting millions of other species. The diversity of bird habitats shows that there are many types of bird species. This study aims to provide a new treasure for the ontology field by applying a machine learning algorithm to classify bird species based on K-Nearest Neighbor (KNN). A total of 400 species of bird images with a total of 58388 images of data were tested in this study. The test scenario was carried out in 3 stages: 400, 200, and 100 species. The results of testing the accuracy of the K-Nearest Neighbor model applied to the bird image dataset are 26.846%, with a value of $\mathrm{K}=1$. The comparison of training data with test data is 95:5 percent (%). This result shows that the KNN algorithm can classify bird species.