Klasifikasi Kematangan Buah Kelapa Sawit Berdasarkan Warna Menggunakan Metode K-Nearest Neighbor

Abstract

The K-Nearest Neighbor (K-NN) algorithm is a simple machine learning algorithm used for classification and regression. This study aims to implement the K-NN algorithm in classifying the ripeness level of oil palm fruit based on color. The data used consisted of 270 images of dura, tenera, and pisifera oil palm fruits taken using a smartphone camera. The results showed that the K value in the K-NN algorithm plays an important role in determining the classification performance. With K = 3, the model achieved the highest accuracy of 93.67%, while the lowest accuracy was 80.05% with a value of K = 25. Compared to previous studies that obtained the highest accuracy of 92% at K = 7, this study shows an increase in classification performance. Classification data analysis showed that 56 image data were correctly classified and 25 image data were incorrectly classified from a total of 81 test image data. This study proves that K-NN with RGB color images can be effectively used for classification of the ripeness level of oil palm fruit.