cover
Contact Name
Mesran
Contact Email
mesran.skom.mkom@gmail.com
Phone
+6282365336853
Journal Mail Official
jurnal.bulletinds@gmail.com
Editorial Address
Jalan sisingamangaraja No 338 Medan, Indonesia
Location
Kota medan,
Sumatera utara
INDONESIA
Bulletin of Data Science
ISSN : -     EISSN : 28079493     DOI : -
Bulletin of Data Science journal publishes manuscripts within the fields of: 1. Soft Computing, 2. Experts System, 3. Decision Support System, 4. Cryptography, 5. Big Data, 6. Data Mining, 7. Artificial Inteligence, and etc
Articles 64 Documents
Evaluasi Prediksi Stres Berdasarkan Tidur dan Kognitif Menggunakan Machine Learning Ibnu Sina Maulana; Hafiyyan Putra Pratama
Bulletin of Data Science Vol 6 No 1 (2026): October 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletinds.v6i1.10128

Abstract

Sleep and cognitive indicators are theoretically relevant to stress, but this relevance does not guarantee individual prediction in small datasets. This exploratory study evaluates the learnability of Stress_Level in the public Sleep Deprivation and Cognitive Performance dataset containing 60 observations and 12 predictors. Descriptive analysis is complemented by nested cross-validation with five outer folds, three repetitions, and three inner folds. Encoding, scaling, and feature selection are fitted exclusively on training data. Six regression families are compared with a training-mean baseline, least absolute shrinkage and selection operator (LASSO), and recursive feature elimination (RFE) with Ridge. All marginal 95% correlation intervals include zero; the largest correlation concerns sleep quality (0.190). All feature-based models yield negative mean coefficients of determination, ranging from −0.470 to −0.235, indicating failure to generalize in this internal evaluation. The baseline achieves the lowest mean absolute error of 9.65 ± 1.33, compared with 9.69 ± 1.35 for RFE–Ridge and 9.76 ± 1.41 for Ridge; values after ± represent standard deviations across outer folds. Averaged cross-validated Ridge predictions concentrate within 14.67–19.72 despite the target range of 0–40. Tuning and feature selection do not demonstrate added predictive value over the baseline. These negative findings diagnose the information limits of this dataset and do not support stress screening or diagnosis. Small sample size, limited instrument documentation, and absent external validation restrict inference. Future studies require validated stress measurements, longitudinal observations, and temporally aligned physiological and psychosocial features.
Analisis Komparatif Kinerja IndoBERT dan XLM-RoBERTa Dalam Analisis Sentimen Ulasan Google Maps Mitra10 Siliwangi Pamulang Rizki Reza; Sajarwo Anggai; Abu Khalid Rivai
Bulletin of Data Science Vol 6 No 1 (2026): October 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletinds.v6i1.11071

Abstract

Google Maps customer reviews provide rich information about shopping experiences, but the growing volume of unstructured text makes manual analysis inefficient. This study classifies customer-review sentiment for Mitra10 Siliwangi Pamulang and compares IndoBERT, an Indonesian monolingual language model, with XLM-RoBERTa, a multilingual model. A total of 1,305 reviews were collected through web scraping and categorized into positive, neutral, and negative sentiment. Data processing included selection, duplicate removal, case folding, text-format normalization, tokenization, a maximum input length of 128 tokens, and dataset splitting into 1,043 training samples, 131 validation samples, and 131 test samples. Class imbalance was handled using class weighting in weighted CrossEntropyLoss without oversampling. Both models were fine-tuned with identical settings: 5 epochs, batch size 8, learning rate 2×10⁻⁵, AdamW optimizer, and random seed 42. XLM-RoBERTa achieved 96.95% accuracy, 97.25% weighted precision, 96.95% weighted recall, and a 97.07% weighted F1-score. IndoBERT achieved 95.42% accuracy, 96.90% weighted precision, 95.42% weighted recall, and a 95.88% weighted F1-score. Although XLM-RoBERTa was numerically 1.53 percentage points higher in accuracy, the exact McNemar test produced a p-value of 0.625, indicating no statistically significant difference on the test set. Both models therefore performed strongly on this dataset; however, the negative class should be interpreted cautiously because only 40 of the 1,305 reviews were negative.
Comparative Machine Learning Approaches for Classifying Student Stress Types Chrisensia Abigail Gani; Clarice Harijanto; Christopher Andreas
Bulletin of Data Science Vol 6 No 1 (2026): October 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletinds.v6i1.11202

Abstract

Student stress is a major mental health concern that may affect academic performance, psychological well-being, and quality of life. Early identification of stress types is important because different stress categories may require different intervention strategies. With advances in data-driven technologies, machine learning has emerged as a promising approach for identifying patterns in mental health data and supporting automated assessment. This study aims to classify student stress types using the Stress Indicators Dataset for Mental Health Classification, which contains 2,000 student responses. The dataset includes demographic information, psychological symptoms, academic challenges, physical health indicators, and social factors associated with stress. Before model development, preprocessing procedures included data cleaning, duplicate removal, filtering extreme age values, and exploratory data analysis (EDA). Feature engineering was also performed by creating three composite variables, namely symptom_score, academic_stress_score, and social_stress_score, to represent broader stress-related dimensions. Three classification algorithms, Logistic Regression, K-Nearest Neighbors (KNN), and Gaussian Naive Bayes, were developed and evaluated using confusion matrix analysis and Accuracy, Precision, Recall, and F1-Score metrics. The dataset was split into training and testing subsets using an 80:20 stratified split to preserve class distribution. Results indicate that Logistic Regression achieved the highest performance, with an accuracy of 98.10%, precision of 98.14%, recall of 98.10%, and F1-score of 97.96%. Feature engineering slightly improved KNN performance but did not enhance Logistic Regression or Gaussian Naive Bayes. These findings demonstrate the potential of machine learning as a supporting tool for student stress assessment and monitoring, enabling identification of stress categories.
A Analisis Sentimen Purbaya Effect pada X Menggunakan VADER Sentiment, SVM, dan SMOTE Taki Imay; Baginda Oloan Lubis; Besus Maula Sulthon
Bulletin of Data Science Vol 6 No 1 (2026): October 2026
Publisher : Forum Kerjasama Pendidikan Tinggi (FKPT)

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47065/bulletinds.v6i1.11264

Abstract

The rapid growth of social media as a platform for public expression has generated a substantial volume of unstructured opinion data, creating the need for computational-based analytical approaches. One phenomenon that attracted public attention on platform X is the Purbaya Effect, which triggered various public responses reflected in positive, neutral, and negative sentiments. This study aims to analyze public sentiment toward this phenomenon by integrating VADER Sentiment, Support Vector Machine (SVM), and Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance in the dataset. This research employed a quantitative approach based on text mining using 2,871 Indonesian-language tweets collected from platform X during the period of October 2025 to May 2026. The research procedure consisted of data collection, preprocessing (cleaning, case folding, tokenization, and stopword removal), sentiment labeling using VADER, data balancing using SMOTE, sentiment classification using SVM, and model evaluation through accuracy, precision, recall, and F1-score metrics. The results indicate that the sentiment distribution was dominated by the neutral category, while positive and negative sentiments appeared in considerably smaller proportions. The SVM model before applying SMOTE achieved an accuracy of 91.80%, but demonstrated a strong bias toward the majority class, resulting in limited capability to identify minority classes. After implementing SMOTE, the dataset became more balanced and the model’s ability to detect minority classes improved, as reflected by increased recall and F1-score values, although the overall accuracy decreased to 77.87%. These findings suggest that applying SMOTE improves the quality of multiclass sentiment classification and provides a more representative evaluation than relying solely on accuracy as a performance indicator.