Ibnurayyan Rayyan
Sekolah Tinggi Manajemen Informatika dan Komputer Kaputama Binjai

Published : 1 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 1 Documents
Search

Sistem Pakar Penilaian Mutu Kopi Menggunakan Metode Bayes Ibnurayyan Rayyan; Mili Alfhi Syari; Muhammad Fauzi
Merkurius : Jurnal Riset Sistem Informasi dan Teknik Informatika Vol. 4 No. 5 (2026): Merkurius : Jurnal Riset Sistem Informasi dan Teknik Informatika
Publisher : Asosiasi Riset Teknik Elektro dan Informatika Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61132/merkurius.v4i5.1825

Abstract

Coffee quality grading requires consistent evaluation of physical and sensory attributes, yet manual assessment may be time-consuming and influenced by assessor subjectivity. This study develops a web-based expert system using the Bayes method to support Arabica coffee quality assessment at Roman Kopi Gayo, Bener Meriah Regency. The study used 100 grading records, divided into 80 training records and 20 testing records. Six criteria were modeled: screen size, bean color, dominant aroma, moisture content, defect count, and cupping score. The system calculates class priors, criterion likelihoods, unnormalized class scores, and normalized posterior probabilities for four quality categories. It was implemented with native PHP, MySQL, HTML, CSS, and JavaScript in an XAMPP 5.6 environment. For the reported sample—screen size 17, 11.1% moisture, three defects, bluish-green beans, caramel-fruity aroma, and a cupping score of 86.75—the system classified the coffee as Very Good Quality with a posterior probability of 98.2364%. Functional verification comprised 24 black-box scenarios, all successful, and 12 white-box logic paths, all conforming to the design. The application therefore provides a traceable and consistent decision-support mechanism for preliminary coffee grading. Model accuracy, however, must still be established by comparing predictions with the actual labels of all testing records.