KLASIFIKASI TUMBUHAN ANGIOSPERMAE MENGGUNAKAN ALGORITMA K-NEAREST NEIGHBOR BERDASARKAN PADA BENTUK DAUN

Abstract

didapatkan sebesar 81% dengan jumlah tetangga k=1. Setelah dilakukan pengujian, akan dilakukan evaluasi model dengan menggunakan confusion matrix, yang terdiri dari accuracy, precision, recall, dan f-1 score. Kata Kunci: angiospermae, canny, confusion matrix, k-nearest neighbor, matriks populasi ABSTRACT About 250,000 living plant species are plants of the Angiosperms group. The diversity of Angiosperm plants in Indonesia makes it difficult for someone to carry out the process of identifying plants into the monocotyledonous or dicotyledonous class. This is the background for doing this research by classifying Angiosperms plants using the K-Nearest Neighbor algorithm. Research related to classifying manga plants using the Backpropagation algorithm with feature extraction using GLCM produces an accuracy of 49%. Furthermore, the classification of plants using the K-Nearest Neighbor algorithm with leaf morphological features extraction resulted in an accuracy of 92%. There is also a classification of citrus plants using the K-Nearest Neighbor algorithm with leaf texture feature extraction resulting in an accuracy of 81.48%. In this study, the K-Nearest Neighbor classification algorithm is used with shape feature extraction using a population matrix with the aim of examining the level of classification accuracy using population matrix feature extraction. The data used is a digital leaf image which will extract its shape features using a population matrix feature by doing image pre-processing and detection first. The data is then divided into training data of 70%, and test data of 30%. The results of this study are the accuracy of the test data obtained by 81% with the number of neighbors k=1. After testing, an evaluation of the model will be carried out using a confusion matrix, which consists of accuracy, precision, recall, and f-1 score.