Implementasi Sistem Pakar Untuk Diagnosis Hama Tanaman Kelapa Sawit Dengan Metode Certainty Factor

This study proposes the development of an expert system based on the Certainty Factor method for rapid and accurate diagnosis of oil palm pests. The system is designed to help farmers identify four main types of pests (rats, termites, caterpillars, and bagworms) based on 18 specific symptoms and provide control recommendations. Data were obtained from interviews with agricultural experts, literature studies, and field observations in a 2.5-hectare oil palm plantation in Tanjung Selor. Users expressed their confidence level using a five-point scale (none, slightly confident, fairly confident, confident, and very confident). The Certainty Factor method calculates the confidence level of the diagnosis by combining the observed symptoms. The system was implemented as an offline desktop application based on Python and PyQt5. Functional testing (Black Box Testing) on ​​14 scenarios showed that all features functioned according to specifications. Performance evaluation included diagnostic accuracy (confusion matrix), Certainty Factor validation, and usability (System Usability Scale). The results showed 100% accuracy, precision, recall, and F1-Score in 20 sample cases. An average SUS score of 85.125 (Acceptable category approaching Excellent) from 40 respondents indicated ease of use. The system has proven effective in supporting decision-making for farmers and extension workers. Further development is recommended for secondary pests, plant diseases, and web and mobile platforms.
Penelitian ini mengusulkan pengembangan sistem pakar berbasis metode Certainty Factor untuk diagnosis hama tanaman kelapa sawit secara cepat dan akurat. Sistem dirancang membantu petani mengidentifikasi empat jenis hama utama (tikus, rayap, ulat api, ulat kantong) berdasarkan 18 gejala spesifik serta memberikan rekomendasi pengendalian. Data diperoleh dari wawancara pakar pertanian, studi literatur, dan observasi lapangan di perkebunan kelapa sawit seluas 2,5 hektar di Tanjung Selor. Pengguna menyatakan tingkat keyakinan melalui lima skala (tidak ada, sedikit yakin, cukup yakin, yakin, sangat yakin). Metode Certainty Factor menghitung tingkat kepercayaan diagnosis dengan mengombinasikan gejala yang diamati. Sistem diimplementasikan sebagai aplikasi desktop offline berbasis Python dan PyQt5. Pengujian fungsional (Black Box Testing) terhadap 14 skenario menunjukkan semua fitur berfungsi sesuai spesifikasi. Evaluasi kinerja mencakup akurasi diagnosis (confusion matrix), validasi Certainty Factor, dan usabilitas (System Usability Scale). Hasil menunjukkan akurasi, presisi, recall, dan F1-Score 100% pada 20 kasus sampel. Skor SUS rata-rata 85,125 (kategori Acceptable mendekati Excellent) dari 40 responden mengindikasikan kemudahan penggunaan. Sistem terbukti efektif sebagai pendukung keputusan petani dan penyuluh. Pengembangan lanjutan disarankan untuk hama sekunder, penyakit tanaman, serta platform web dan mobile.