Shafril Dwi Firmansyah
Universitas Papua

Published : 2 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 2 Documents
Search

Analisis Sentimen Program Makan Bergizi Gratis Siswa SMAN 01 Manokwari dengan Naïve Bayes Ridho Nur Fauzi; Thoriq N.F Wajo; Shafril Dwi Firmansyah; Reason Dominik Harewan; Iqbal Dwi Wijanarko
Jurnal Ilmiah Binary STMIK Bina Nusantara Jaya Lubuklinggau Vol 7 No 2 (2025): Jurnal Ilmiah Binary STMIK Bina Nusantara Jaya
Publisher : STMIK Bina Nusantara Jaya Lubuk Linggau

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.52303/jb.v7i2.161

Abstract

Abstract The research aims to analyze the sentiment of students at SMAN 01 Manokwari toward the effectiveness of the Nutritious Meal Program (MBG), using one of the classification methods, namely the Naïve Bayes method. A total of 252 data points were analyzed, originating from student opinions that had undergone manual labeling into three sentiment categories: positive, neutral, and negative. The data distribution showed that 74.2% of the comments were positive, 15.1% neutral, and 10.7% negative. The data was then split into 80% training data and 20% test data. Before classification, feature extraction was performed using the TF-IDF method. The classification results showed that the Naïve Bayes algorithm was able to classify the data with a fairly high accuracy of 86.27%. The highest precision was obtained in the positive class at 0.92. Likewise, the highest recall was also achieved in the positive class at 0.97. The highest F1-Score was also obtained in the positive class, reaching 0.95, and the AUC value reached 0.98, which falls into the category of excellent classification. Based on these results, it can be concluded that the majority of students gave positive responses to the Free Nutritious Meal Program, and the Naïve Bayes method is quite effective for sentiment analysis of student opinion data. Abstrak Penelitian bertujuan untuk menganalisis sentimen pada siswa SMAN 01 Manokwari terhadap efektivitas Program Makan Bergizi (MBG) pada siswa, dengan menggunakan salah satu metode klasifikasi yaitu metode Naïve Bayes. Jumlah Data yang telah dianalisis sebanyak 252 yang berasal dari pendapat para siswa yang telah melalui proses pelabelan manual menjadi tiga kategori sentimen, yaitu positif, netral, dan negatif. Distribusi pada data menunjukkan 74,2% komentar ber sentimen positif, 15,1% netral, dan 10,7% negatif. Data kemudian dibagi menjadi 80% data latih dan 20% data uji sebelum dilakukan pengekstrakan fitur pada data, maka fitur menggunakan metode TF-IDF. Hasil dari klasifikasikasi menunjukkan bahwa algoritma Naïve Bayes dapat mengklasifikasikan data dengan akurasi yang cukup besar 86,27%. Precision tertinggi yang telah diperoleh pada kelas positif sebesar 0,92. Sedangkan untuk recall tertinggi juga diperoleh pada kelas positif sebesar 0,97. Nilai F1-Score tertinggi juga didapatkan oleh data positif yang mana untuk kelas positif mencapai 0,95 dan nilai AUC sebesar 0,98 yang termasuk dalam kategori excellent classification. Berdasarkan hasil tersebut, dapat disimpulkan bahwa sebagian besar siswa memberikan tanggapan positif terhadap program Makan Bergizi Gratis, dan metode Naïve Bayes cukup efektif digunakan dalam analisis sentimen data pada pendapat siswa.
Design and Functional Evaluation of an Eastern Indonesia-Themed Arithmetic Learning Game Using Construct 2 Shafril Dwi Firmansyah; Julius Panda Putra Naibaho; Alex De Kweldju
EDUTIC Vol 13, No 2: 2026
Publisher : Universitas Trunodjoyo Madura

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.21107/edutic.v13i2.35073

Abstract

Basic arithmetic mastery is an essential part of early numeracy, yet learners require interactive and repeated practice to understand addition, subtraction, multiplication, and division. This study developed and evaluated the core functions of Petualangan Matematika, an Eastern Indonesia-themed arithmetic educational game created using Construct 2. A prototype-development approach was employed, comprising needs analysis, interface and flow design, asset and game-logic development, and black-box testing. The HTML5 prototype was tested in a Chromium-based browser at a 1280 × 720-pixel landscape resolution. Fourteen test scenarios covered application loading, navigation, audio control, presentation of learning materials, correct and incorrect answer responses, question transitions, and score calculation. All core functions operated according to the expected flow. These findings demonstrate initial functional feasibility but do not establish content validity, usability, user satisfaction, or learning effectiveness. Further research should involve expert validation, user testing, expansion of the question bank, corrective feedback, and evaluation of learning outcomes.