ADELINA LOPO Pengarang WENEFRIDA T INA Dosen Pembimbing 1 MOLINA OLIVIA ODJA Dosen Pembimbing 2 Wenefrida T Ina Ketua Penguji Molina Olivia Odja Penguji 1 Hendrik J Djahi Penguji 2

K-Nearest Neighbors (K-NN) is one of the simplest machine learning algorithms and is widely used in classification and regression. The way KNN works is approaching or the closest distance between training data and test data. This study aims to determine the classification of lime fruit into four levels, namely raw oranges, moderately ripe, ripe oranges and rotten oranges. The research method used is the study of literature, design and implementation of a classification program using the MATLAB GUI. The results of research conducted using K values = 1, 3, 5, 7 and 9, odd K values were chosen to avoid the ease of distance which can make the classification results inaccurate. The training data used as many as 40 images and test data as many as 20 images with dimensions of 512 x 512, with type .jpg. The results for the raw orange class have an accuracy value of 86% (k = 7 and 9), for the slightly ripe orange class the accuracy value is 87% (k = 9), for the ripe orange class and rotten orange both have an 87% accuracy value. at the same value of k, namely (k = 5, 7, and 9). Keywords: KNN, fruit maturity level, lime, fruit skin color