cover
Article Per Year (5 Year)
All
Home Page
OAI Link
Editorial Team
Contact
Reviewer
Google Scholar
Contact Name
Muhammad Syahrizal
Contact Email
syahrizal83.budidarma@gmail.com
Phone
+6282370070808
Journal Mail Official
pdsi.bids@gmail.com
Editorial Address
Jalan sisingamangaraja No 338 Medan, Indonesia
Location
Kota medan,
Sumatera utara
INDONESIA
Bulletin of Informatics and Data Science
Published by Asosiasi Peneliti Data Science Indonesia
ISSN : -     EISSN : 25808389     DOI : -
Core Subject : Science, Engineering,
Computer Science & IT Electrical & Electronics Engineering Engineering
The Bulletin of Informatics and Data Science journal discusses studies in the fields of Informatics, DSS, AI, and ES, as a forum for expressing research results both conceptually and technically related to Data Science
Arjuna Subject : Ilmu Komputer (semua kategori) - Kecerdasan Buatan
Articles 54 Documents
 2   3   4   5   6 
Waste Classification using EfficientNetB3-Based Deep Learning for Supporting Sustainable Waste Management Agustiani, Sarifah; Junaidi, Agus; Aryanti, Riska; Kamil, Anton Abdul Basah
Bulletin of Informatics and Data Science Vol 4, No 1 (2025): May 2025
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v4i1.108

Abstract

Waste management is a critical issue in sustainable development, particularly in large urban areas that generate a high volume of waste daily. One of the main challenges is the absence of a fast, accurate, and efficient waste sorting system. This study aims to develop a waste classification model using deep learning based on the EfficientNetB3 architecture to support more sustainable waste management. The model was trained on a dataset obtained from a Kaggle repository, consisting of 4,650 images evenly distributed across six waste categories: batteries, glass, metal, organic, paper, and plastic (775 images per class). The training and evaluation were conducted using a supervised image classification approach. The model achieved an overall accuracy of 93%, with average precision, recall, and F1-score values of 93%. Among all categories, organic waste achieved the highest F1-score (0.99), followed by paper (0.97) and batteries (0.97), while plastic and metal categories obtained F1-scores of 0.89. These results demonstrate that the EfficientNetB3 architecture is effective in performing multi-class waste classification. This model has the potential to be implemented in camera-based waste sorting systems such as smart bins or automated recycling units, thereby contributing to the reduction of unprocessed waste and supporting the achievement of Sustainable Development Goal (SDG) 12: responsible consumption and production
Weighted Multi-Criteria Assessment of Rice Quality Using The TOPSIS Method Satria, Budy; Fadilah, Sandi
Bulletin of Informatics and Data Science Vol 4, No 2 (2025): November 2025
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v4i2.145

Abstract

Rice is a staple food for the Indonesian people, and its availability must be guaranteed by the government. The background of this research is based on the increasing demand for high-quality rice from consumers, thus challenging producers to set optimal rice quality standards. The process of selecting quality rice is still carried out using conventional methods in Bulog warehouses, namely by checking every rice data received by the quality control team tasked with assessing the quality of incoming rice. To overcome this problem, a decision support system is needed that can provide fair, objective, and efficient decisions. This study aims to evaluate the quality of rice from 10 alternatives using five criteria: milling degree, head grain, moisture content, broken grain, and grit grain, with a total weight of 100%. The Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method is applied. This research was conducted by following a series of steps, including building a Decision Matrix, Normalizing the Decision Matrix, Calculating the Weighted Normalized Decision Matrix, Determining the Ideal Positive and Negative Solutions, Calculating the Distance to the Ideal Positive and Negative Solutions, and Calculating the Preference Score. The results of the study showed that from 10 alternative data, 5 types of rice were obtained with the highest preference values, namely Harum Solok Rice (0.8363), Anak Daro Rice (0.7955), Kuruik Kusuik Rice (0.7210), Ampek Angkek Rice (0.6919), and Saganggam Panuah Rice (0.6727). The conclusion of this study is that the application of the TOPSIS method is effective in objectively assessing rice quality. In further research, it is recommended to utilize a combination of other decision support methods to acquire new knowledge and refine preference values, as well as to develop these methods into user-friendly interfaces
Hybrid Chaos-Isolation Forest Framework for Anomaly Detection in Indonesia’s Public Procurement Ambarsari, Erlin Windia; Desyanti, Desyanti; Fathudin, Dedin
Bulletin of Informatics and Data Science Vol 4, No 2 (2025): November 2025
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v4i2.137

Abstract

This study proposes and empirically evaluates a Hybrid Chaos-Isolation Forest (HC-iForest) framework for detecting anomalies in Indonesia’s public procurement datasets. The purpose of this research is to address the difficulty of identifying irregular procurement patterns, as existing assessment mechanisms remain largely descriptive and retrospective. The framework integrates chaos-based temporal descriptors—permutation entropy, turning points, and volatility—with statistical indicators to enhance sensitivity to nonlinear and irregular time series. Using monthly procurement data from the Open Contracting Data Standard (OCDS) covering the period from 2019 to 2024, the model identified anomalous fiscal patterns associated with year-end budget adjustments and procurement surges. Empirical evaluation using correlation, ablation, and statistical validation shows that the hybrid model introduces non-redundant anomaly information, achieving a Spearman rank correlation of approximately 0.75 compared to the baseline Isolation Forest, with reduced overlap at intermediate thresholds (Jaccard similarity of 0.20 at the Top 5%). These results confirm that chaos-driven features improve model stability and interpretability. The findings reveal that anomalies are systemic manifestations of institutional and fiscal behavior rather than random deviations. The HC-iForest framework offers a data-driven early-warning mechanism for oversight agencies such as LKPP and ICW, strengthening transparency and accountability in public spending. Future studies may extend this framework through neural or spatiotemporal hybrid architectures to support intelligent and adaptive fiscal monitoring systems
Classification Model Optimization using Grid Search and Random Search in Machine Learning Algorithms Parinduri, Syawaluddin Kadafi; Alkhairi, Putrama; Irawan, Irawan; Qurniawan, Hendry
Bulletin of Informatics and Data Science Vol 4, No 2 (2025): November 2025
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v4i2.136

Abstract

The performance of a machine learning model is highly dependent on the selection and tuning of appropriate hyperparameters. The main problem in this study is how to improve the accuracy and stability of a classification model without sacrificing computational time efficiency, especially in the case of kidney disease classification that requires accurate and fast prediction results. This study aims to optimize the classification model by applying two hyperparameter search methods, namely Grid Search and Random Search, to the Random Forest algorithm. The kidney disease dataset is used as a case study with preprocessing processes including data cleaning, missing value imputation, categorical variable encoding, and normalization. Each model is tested using accuracy, precision, recall, and F1-Score metrics. The results show that the Grid Search_RF model produces the highest performance with perfect accuracy, precision, recall, and F1-Score values (1.0000), while Random Search_RF provides results close to (accuracy 0.9875 and F1-Score 0.9900) with more efficient training time. Meanwhile, the standard Random Forest without tuning still shows competitive performance (accuracy 0.9917 and F1-Score 0.9930). Based on these results, it can be concluded that hyperparameter optimization, using both Grid Search and Random Search, can significantly improve the performance of the classification model, with Random Search being the most efficient method for practical implementation in machine learning-based disease detection systems.
Comparison of Case-Based Reasoning and Hybrid Case-Based Methods in Expert System for Diagnosing Rice Plant Diseases Roznim, Roznim; Mesran, M.Kom, Mesran; Setiawansyah, Setiawansyah; Ambarsari, Erlin Windia
Bulletin of Informatics and Data Science Vol 4, No 2 (2025): November 2025
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v4i2.132

Abstract

Rice plants are susceptible to various types of diseases that can reduce productivity and quality of the harvest. Therefore, an expert system is needed that can help the disease diagnosis process quickly and accurately. This study compares two approaches in expert systems, namely the Case-Based Reasoning (CBR) method and the Hybrid Case-Based method, to diagnose rice plant diseases based on the symptoms experienced. Data on symptoms and types of diseases were analyzed using both methods to see the level of suitability of the resulting diagnosis. The test results showed that the Hybrid Case-Based method produced a higher level of certainty for all types of diseases compared to the CBR method. For example, Bacterial Leaf Blight disease has a certainty value of 99.5% in the Hybrid method, higher than 83.8% in the CBR method. These findings indicate that the Hybrid method is more effective and accurate in the process of diagnosing rice plant diseases. Thus, an expert system based on the Hybrid Case-Based method is recommended to support decision making in the agricultural sector, especially in early detection of rice diseases
Deep Learning–Based Pneumonia Classification on Chest X-Ray Images Mustakim Mustakim; Danur Lestari; Hartono Hartono
Bulletin of Informatics and Data Science Vol 4, No 2 (2025): November 2025
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v4i2.130

Abstract

Pneumonia is a lung infection and one of the leading causes of mortality worldwide. Early and accurate diagnosis is essential to reduce death rates, with chest X-ray (CXR) imaging being the most commonly used diagnostic tool. However, CXR-based pneumonia identification remains challenging due to limited image quality and the shortage of experienced radiologists. To address this issue, this study proposes a hybrid deep learning framework that integrates Convolutional Neural Network (CNN), Recurrent Neural Network (RNN), and Generative Adversarial Network (GAN) to enhance the classification of bacterial and viral pneumonia from CXR images. The dataset comprises 7,927 CXR images, including 3,270 normal cases, 3,001 cases of bacterial pneumonia, and 1,656 cases of viral pneumonia. Four CNN architectures, Xception, InceptionV3, ResNet50V2, and DenseNet201, are evaluated using RMSprop and Stochastic Gradient Descent (SGD) optimizers. Model development and training are conducted using the TensorFlow framework. Experimental results demonstrate that ResNet50V2 with the RMSprop optimizer achieves the highest classification accuracy of 0.85, while also yielding the fastest training time of 2,215 seconds. These findings indicate that the proposed approach can support faster and more accurate pneumonia screening, particularly in healthcare facilities with limited diagnostic resources
Explainable Machine Learning for Multi-Class Classification of Internet Firewall Traffic Titik Misriati; Riska Aryanti
Bulletin of Informatics and Data Science Vol 5, No 1 (2026): May 2026
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v5i1.163

Abstract

The increasing diversity and scale of network traffic introduce significant challenges in performing accurate and interpretable firewall analysis. This research aims to bridge the gap between predictive performance and model transparency by developing an explainable machine learning framework for multi-class firewall traffic classification. The study utilizes the Internet Firewall Data dataset consisting of 65,532 network traffic instances distributed across four firewall action classes and evaluates seven classification algorithms, including Decision Tree, Random Forest, XGBoost, Support Vector Machine, k-Nearest Neighbors, Naïve Bayes, and Logistic Regression. The dataset was partitioned using a stratified 80:20 hold-out approach to preserve the original class distribution and the experimental process involves data preprocessing, normalization, and validation on an independent test set using accuracy, precision, recall, and F1-score metrics. The findings reveal that XGBoost achieves the highest performance, reaching an accuracy of 99.81%, followed by Decision Tree and Random Forest. This indicates that ensemble and tree-based approaches are highly effective in modeling complex and non-linear traffic patterns. To improve interpretability, this study incorporates explainable artificial intelligence techniques, including feature importance and SHAP analysis. The results show that traffic-related attributes significantly influence classification outcomes, providing meaningful insights into firewall decision behavior
Comparative Analysis of MCDM Methods in Employee Award Ranking Roznim binti Mohamad Rasli; Mesran Mesran; Ridha Maya Faza Lubis
Bulletin of Informatics and Data Science Vol 5, No 1 (2026): May 2026
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v5i1.141

Abstract

Awards are an important form of appreciation for outstanding employees, but the process of selecting recipients is often faced with various challenges, especially in assessing subjective aspects of performance. To overcome this, the Multi-Criteria Decision Making (MCDM) method can be an effective solution. MCDM offers a systematic framework for evaluating various alternatives (in this case, employees) based on a number of relevant criteria. By using the MCDM method, the process of selecting award recipients can be carried out more objectively and transparently. Some commonly used MCDM methods, such as MAUT, OCRA, and CoCoSo, have their own advantages and disadvantages. This study aims to compare the three methods specifically in the context of selecting employee award recipients. The final results obtained show that the best alternative is A6, where the results of the three methods look the same position or location in the ranking. After a comparative analysis of the three methods, it can be concluded that the OCRA method is the best method in terms of ranking consistency compared to the other two methods. Thus, it is hoped that recommendations for the most suitable MCDM method can be obtained to be applied in similar situations
An Optimized Balanced-Learning Framework for Malignant Skin Lesion Triage Using Compound-Scaled Neural Networks Argha Orion Silitonga; Raissa Camilla Maringka; Wilsen Grivin Mokodaser; George M W Tangka; Marchel Timothy Tombeng
Bulletin of Informatics and Data Science Vol 5, No 1 (2026): May 2026
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v5i1.165

Abstract

Skin cancer represents a prevalent global health challenge, and early detection is very important to reduce mortality risk. Manual dermoscopic diagnosis risks human bias, making deep learning classification a vital research topic. While several previous studies utilizing the ISIC 2019 dataset have demonstrated high diagnostic capabilities, they primarily focus on complex multi-class classification. However, in real-world clinical workflows, the primary necessity is a swift, dependable triage system that can confidently distinguish dangerous lesions from non-threatening ones. Furthermore, many existing models require substantial computational overhead yet still suffer from imbalanced accuracy when dealing with minority malignant classes. The novelty of this study lies in addressing these gaps by developing a streamlined, clinically practical binary screening framework optimized specifically for malignant-versus-benign triage. The original multi-class labels were transformed into binary classes where malignant lesions consist of melanoma (MEL), basal cell carcinoma (BCC), and squamous cell carcinoma (SCC), while benign lesions consist of nevus (NV), benign keratosis (BKL), dermatofibroma (DF), and vascular lesions (VASC). The experiment applied transfer learning with ImageNet-pretrained weights, data augmentation, class weighting, and fourfold stratified cross-validation. Unlike prior works that rely on resource-heavy architectures, we leverage the compound-scaled EfficientNet-B4 backbone—delivering superior feature representational power with significantly fewer parameters evaluate on a large-scale cohort of 25,331 dermoscopic images. Experimental results show that the proposed model achieved an average accuracy of 89.77% and an average ROC AUC of 96.16%. The best fold obtained 91.49% accuracy with ROC AUC of 97.19%. Simultaneously, the framework maintained an average F1-score of 89.20%
Implementation of Convolutional Neural Network (CNN) MobileNetV2 in Lung Disease Classification from X-Ray Images Mohammad Faris Fawwaz; Arif Aryaguna Nauli; Roslina Roslina
Bulletin of Informatics and Data Science Vol 5, No 1 (2026): May 2026
Publisher : PDSI

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.61944/bids.v5i1.131

Abstract

The classification of lung diseases from X-ray images is often challenged by significant data imbalance, where minority classes like COVID-19 constitute only approximately 20% of the dataset compared to the majority classes. This condition can degrade model performance and introduce bias. This study aims to analyze the impact of data balancing strategies and training parameter variations to improve the accuracy of a Convolutional Neural Network (CNN) model based on the MobileNetV2 architecture. The experimental process systematically compared two learning rates (1e-3 and 1e-4) and two optimizers (Adam and RMSprop) across four distinct data handling scenarios: no augmentation, geometric augmentation only, the Mixup technique only, and a combination of both. The model was evaluated on a four-class X-ray image dataset comprising COVID-19, Normal, Pneumonia, and Tuberculosis. The optimal results were achieved by applying the combined approach of geometric augmentation and Mixup with a 1e-3 learning rate and the Adam optimizer. This configuration significantly outperformed other scenarios, reaching a testing accuracy of 96.62% and an average F1-Score of 96.63%, demonstrating excellent model generalization. This high-performing model has been successfully implemented in a mobile application using Flutter and TensorFlow Lite, serving as a practical tool to support the early diagnosis of lung diseases

Page 5 of 6 | Total Record : 54

 2   3   4   5   6 

Filter by Year

2022 2026


Reset
Filter By Issues
All Issue Vol 5, No 1 (2026): May 2026 Vol 4, No 2 (2025): November 2025 Vol 4, No 1 (2025): May 2025 Vol 3, No 2 (2024): November 2024 Vol 3, No 1 (2024): May 2024 Vol 2, No 2 (2023): November 2023 Vol 2, No 1 (2023): May 2023 Vol 1, No 2 (2022): November 2022 Vol 1, No 1 (2022): May 2022