Comparison Of K-nearest Neighbor (KNN) And Linear Discriminant Analysis (LDA) Algorithms For Mature Ajwa Date Fruit Classification

Abstract

Currently, many applications of artificial intelligence in various fields of life, especially in image data, require digital image processing. One example of the use of digital images often encountered is image processing of fruit ripeness. Dates are a fruit in great demand by the people of Indonesia, and one of the most popular dates is the Ajwa date. The author is interested in developing previous research regarding identifying the ripeness of Ajwa Dates, where previous research used RGB color image processing with the HIS method. Therefore, the authors want to apply a different method, namely the K-Nearest Neighbor (K-NN) method and Linear Discriminant Analysis (LDA), in classifying the ripeness of the Ajwa Dates by applying a statistical feature algorithm. This research aims to develop a classification model for the maturity level of Ajwa Dates. Furthermore, it is expected to provide better classification results than the previous algorithm. The test results using the KNN method can produce higher accuracy than the LDA, where the KNN method is obtained from the calculation of the Euclidean distance k = 1 100% and Manhattan with a value of k = 1 and k = 2 worth 100%, but the minimum accuracy value is 53.33 % is found at k = 9 in the Euclidean distance calculation, while the LDA accuracy value can reach 93.33%.