Analisis Sentimen Mahasiswa STMIK Widya Cipta Dharma Terhadap Penggunaan Welearn Menggunakan Algoritma Naïve Bayes

Hidayatullah, Ramadhini (2026) Analisis Sentimen Mahasiswa STMIK Widya Cipta Dharma Terhadap Penggunaan Welearn Menggunakan Algoritma Naïve Bayes. S1 Teknik Informatika thesis, STMIK Widya Cipta Dharma.

	Text 2243064-S1-Jurnal.pdf Download (504kB)
	Text 2243064-S1-Teknik Informatika.pdf Restricted to Repository staff only Download (2MB) | Request a copy

Abstract

Perkembangan teknologi informasi mendorong pemanfaatan Learning Management System (LMS) sebagai media pembelajaran di perguruan tinggi. STMIK Widya Cipta Dharma menggunakan WeLearn sebagai platform pembelajaran digital untuk mendukung proses perkuliahan, presensi, pengumpulan tugas, dan pelaksanaan ujian. Beragam tanggapan mahasiswa terhadap penggunaan WeLearn perlu dianalisis untuk mengetahui persepsi mahasiswa sekaligus menjadi bahan evaluasi dalam meningkatkan kualitas layanan. Oleh karena itu, penelitian ini bertujuan untuk menganalisis sentimen mahasiswa STMIK Widya Cipta Dharma terhadap penggunaan WeLearn menggunakan algoritma Naïve Bayes. Penelitian ini menggunakan metode Cross Industry Standard Process for Data Mining (CRISP-DM) yang meliputi tahapan Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, dan Deployment. Data penelitian diperoleh melalui penyebaran kuesioner kepada mahasiswa aktif STMIK Widya Cipta Dharma angkatan 2022–2025 sebanyak 168 komentar. Tahap data preparation meliputi case folding, cleaning, tokenizing, filtering, stemming, dan labeling. Pelabelan dilakukan menggunakan tiga teknik, yaitu pelabelan manual, pelabelan menggunakan InSet Lexicon, dan pelabelan menggunakan Artificial Intelligence (AI) dengan ChatGPT. Proses klasifikasi menerapkan varian Complement Naïve Bayes sebagai implementasi algoritma Naïve Bayes, serta teknik RandomOverSampler untuk mengatasi ketidakseimbangan jumlah data pada setiap kelas sentimen. Hasil penelitian menunjukkan bahwa pelabelan manual menghasilkan 124 komentar positif (73,81%), 26 komentar negatif (15,48%), dan 18 komentar netral (10,71%). Pelabelan menggunakan InSet Lexicon menghasilkan 79 komentar positif (47,02%), 47 komentar netral (27,98%), dan 42 komentar negatif (25,00%), sedangkan pelabelan menggunakan AI dengan ChatGPT menghasilkan 120 komentar positif (71,43%), 32 komentar netral (19,05%), dan 16 komentar negatif (9,52%). Hasil evaluasi menunjukkan bahwa model dengan pelabelan manual memperoleh akurasi sebesar 0,82, model dengan pelabelan AI menggunakan ChatGPT memperoleh akurasi sebesar 0,74, sedangkan model dengan pelabelan InSet Lexicon memperoleh akurasi sebesar 0,52. Oleh karena itu, model dengan pelabelan manual dipilih sebagai model terbaik dan diimplementasikan pada website analisis sentimen. Hasil penelitian menunjukkan bahwa algoritma Naïve Bayes mampu mengklasifikasikan sentimen mahasiswa terhadap penggunaan WeLearn sehingga dapat menjadi bahan evaluasi dalam pengembangan layanan pembelajaran digital di STMIK Widya Cipta Dharma. ============================================================= The advancement of information technology has encouraged the adoption of Learning Management Systems (LMS) as learning media in higher education. STMIK Widya Cipta Dharma utilizes WeLearn as a digital learning platform to support lectures, attendance, assignment submission, and examinations. Students' responses to the use of WeLearn need to be analyzed to understand their perceptions and provide evaluation materials for improving service quality. Therefore, this study aims to analyze the sentiment of STMIK Widya Cipta Dharma students toward the use of WeLearn using the Naïve Bayes algorithm. This study employed the Cross Industry Standard Process for Data Mining (CRISP-DM) methodology, which consists of the stages of Business Understanding, Data Understanding, Data Preparation, Modeling, Evaluation, and Deployment. The research data were collected through questionnaires distributed to active students of STMIK Widya Cipta Dharma from the 2022–2025 cohorts, resulting in 168 comments. The data preparation stage included case folding, cleaning, tokenizing, filtering, stemming, and labeling. Labeling was performed using three techniques: manual labeling, InSet Lexicon-based labeling, and AI-assisted labeling using ChatGPT. The classification process applied the Complement Naïve Bayes variant as an implementation of the Naïve Bayes algorithm, along with the RandomOverSampler technique to address class imbalance. The results showed that manual labeling produced 124 positive comments (73.81%), 26 negative comments (15.48%), and 18 neutral comments (10.71%). InSet Lexicon-based labeling produced 79 positive comments (47.02%), 47 neutral comments (27.98%), and 42 negative comments (25.00%), while AI-assisted labeling using ChatGPT produced 120 positive comments (71.43%), 32 neutral comments (19.05%), and 16 negative comments (9.52%). The evaluation results showed that the model built using manual labeling achieved an accuracy of 0.82, the model using AI-assisted labeling with ChatGPT achieved an accuracy of 0.74, while the model using InSet Lexicon-based labeling achieved an accuracy of 0.52. Therefore, the model based on manual labeling was selected as the best-performing model and implemented in the sentiment analysis website. The results indicate that the Naïve Bayes algorithm is capable of classifying students' sentiments toward the use of WeLearn and can serve as an evaluation tool for improving digital learning services at STMIK Widya Cipta Dharma.

Item Type:	Thesis (S1 Teknik Informatika)
Additional Information:	Pembimbing 1 : Ita Arfyanti, S.Kom., M.M Pembimbing 2 : Wahyuni, S.Kom. M.Kom
Uncontrolled Keywords:	Analisis Sentimen, Learning Management System, Naïve Bayes, CRISP-DM
Subjects:	Q Science > QA Mathematics > QA76 Computer software
Divisions:	Teknik Informatika
Depositing User:	Ms Ramadhini Hidayatullah
Date Deposited:	06 Aug 2026 05:48
Last Modified:	06 Aug 2026 05:48
URI:	http://repository.wicida.ac.id/id/eprint/6375

Actions (login required)

View Item