ANALISIS PERBANDINGAN METODE K-NEAREST NEIGHBOR (KNN) dan SUPPORT VECTOR MACHINE (SVM) DALAM DETEKSI PENYAKIT TANAMAN PADI

RIFQI, MUHAMMAD and abdurrasyid, abdurrasyid and Karmila, Sely (2025) ANALISIS PERBANDINGAN METODE K-NEAREST NEIGHBOR (KNN) dan SUPPORT VECTOR MACHINE (SVM) DALAM DETEKSI PENYAKIT TANAMAN PADI. Diploma thesis, ITPLN.

[thumbnail of 202131106_Muhammad Rifqi_Revisi_Skripsi_Muhammad Rifqi.pdf] Text
202131106_Muhammad Rifqi_Revisi_Skripsi_Muhammad Rifqi.pdf
Restricted to Registered users only

Download (6MB)

Abstract

Penyakit pada daun padi dapat mengganggu pertumbuhan tanaman dan berpotensi menurunkan produktivitas. Oleh karena itu, pendekatan berbasis teknologi informasi menjadi salah satu upaya yang relevan untuk mendukung proses deteksi secara lebih terstruktur. Penelitian ini bertujuan membangun model klasifikasi penyakit daun padi menggunakan dua algoritma machine learning, yaitu Support Vector Machine (SVM) dan K-Nearest Neighbor (KNN). Dataset yang digunakan terdiri dari citra daun padi yang terbagi dalam empat kelas: Blast, Blight, Tungro, dan Normal, masing-masing terdiri dari 80 citra. Setiap citra diproses melalui tahapan resize, konversi ke grayscale, dan filtering. Fitur yang diekstraksi meliputi warna (menggunakan histogram), tekstur (menggunakan GLCM), dan bentuk (menggunakan Sobel edge detection). Selanjutnya, dilakukan pembagian data secara stratified 90:10, 80:20, 70:30, 60:40 serta tuning hyperparameter menggunakan GridSearchCV. Evaluasi model dilakukan menggunakan metrik akurasi, precision, recall, F1-score, serta confusion matrix. Hasil menunjukkan bahwa SVM memberikan akurasi sebesar 83% dan F1-score 0,83, sementara KNN memperoleh akurasi 53% dan F1-score 0,53. Temuan ini mengindikasikan bahwa pemilihan algoritma dan parameter yang sesuai, serta kualitas fitur yang diolah, berpengaruh terhadap performa klasifikasi multi-kelas. Implikasi dari penelitian ini diharapkan dapat mendukung pengembangan sistem klasifikasi penyakit padi berbasis citra digital untuk digunakan dalam praktik pertanian presisi.

Diseases affecting rice leaves can disrupt plant growth and potentially reduce productivity. Therefore, an information technology-based approach is one relevant effort to support a more structured detection process. This study aims to develop a classification model for rice leaf diseases using two machine learning algorithms: Support Vector Machine (SVM) and K-Nearest Neighbor (KNN). The dataset consists of rice leaf images divided into four classes: Blast, Blight, Tungro, and Normal, each containing 80 images. Each image was processed through the steps of resizing, conversion to grayscale, and filtering. The extracted features included color (using histogram), texture (using GLCM), and shape (using Sobel edge detection). Next, the data was stratified into an 90:10, 80:20, 70:30, 60:40 split, and hyperparameter tuning was performed using GridSearchCV. Model evaluation was conducted using accuracy, precision, recall, F1-score, and confusion matrix metrics. The results show that SVM provides an accuracy of 83% and an F1-score of 0.83, while KNN achieves an accuracy of 53% and an F1-score of 0.53. These findings indicate that the selection of appropriate algorithms and parameters, as well as the quality of the processed features, influence the performance of multi-class classification. The implications of this research are expected to support the development of a digital image-based rice disease classification system for use in precision agriculture practices.

Item Type: Thesis (Diploma)
Uncontrolled Keywords: daun padi, ekstraksi fitur, klasifikasi citra, K-Nearest Neighbor, Support Vector Machine. rice leaves, feature extraction, image classification, K-Nearest Neighbor, Support Vector Machine.
Subjects: Skripsi
Bidang Keilmuan > Teknik Informatika
Divisions: Fakultas Telematika Energi > S1 Teknik Informatika
Depositing User: Sudarman
Date Deposited: 10 Oct 2025 07:55
Last Modified: 10 Oct 2025 07:55
URI: https://repository.itpln.ac.id/id/eprint/2053

Actions (login required)

View Item
View Item