IJCCS (Indonesian Journal of Computing and Cybernetics Systems)
Not a member yet
480 research outputs found
Sort by
Utilizing Machine Learning for Pattern Recognition of Wayang Kamasan in Efforts to Digitize Traditional Balinese Art
Full text linkThe extinction of local cultural identities gives rise to profound inquiries concerning the conservative approach that may be adopted by a range of stakeholders. The ongoing process of globalization continues to drive technological innovation, while local cultural knowledge is increasingly marginalized. Conversely, an affirmative attitude towards the preservation of local culture is positively correlated with knowledge of local culture. This study focuses on Wayang Kamasan culture and employs a machine learning-based approach to reintroduce Wayang Kamasan in the context of a global community. The research employs a combination of qualitative and experimental quantitative methods. The former is used to gain an in-depth understanding of the socio-cultural aspects of Wayang Kamasan, while the latter are employed to assess the effectiveness of machine learning methods. The findings demonstrate that the machine learning approach to classifying Wayang Kamasan is an effective method for preserving Balinese culture. By accurately classifying the visual identity of Wayang Kamasan, it is possible to digitally document it, thereby facilitating the preservation of Balinese local culture. Pattern recognition through classification enables the preservation of this cultural heritage in digital form while also supporting the recognition of Balinese wayang.
Sentiment Analysis of X Platform on Viral 'Fufufafa' Account Issue in Indonesia Using SVM
Full text linkIn this study, we conducted a comprehensive sentiment analysis of users on the social media platform X concerning the viral controversy surrounding the KasKus account known as “Fufufafa.” This issue attracted widespread attention and sparked varied reactions within the online community. To gain insights into public opinion on the topic, we utilized the Support Vector Machine (SVM) method, a widely recognized machine learning algorithm for classification tasks. The data for this research was gathered from various posts, comments, and public discussions on platform X, which were pre-processed to filter out irrelevant information, such as spam, unrelated topics, and non-informative content. After cleaning the data, user sentiments were categorized into three primary classes: positive, negative, and neutral. The SVM model was then trained and tested using a labeled dataset to accurately predict user sentiments based on the textual content of their interactions
Development and Validation of a Virtual Reality Circumcision Training Simulator: Simulator Sickness, User Experience, and Clinical Performance in Bali, Indonesia
Full text linkVirtual Reality (VR) is increasingly integrated into medical education, yet its application in Indonesia remains limited. This study developed and validated a VR-based circumcision simulator to evaluate simulator sickness, user experience, and clinical performance. A mixed-methods, repeated-measures design was conducted with 74 participants (25 Novices, 24 Intermediates, 25 Experts). Participants engaged in three simulation modes (Autonomous, Guided, Haptic). Instruments included SSQ, FMS, VRNQ, UEQ-S, Checklist, and OSATS. Analyses employed repeated-measures ANOVA, nonparametric tests, and Spearman correlations. Simulator sickness was highest in Autonomous Mode. User experience scores improved with expertise, showing positive correlations with performance and negative correlations with sickness. Experts consistently outperformed other groups, and skill improvements were retained for up to one month. The VR circumcision simulator demonstrated strong construct validity and educational impact. Instructional modes effectively reduced sickness, while haptic integration enhanced spatial orientation. Future studies should incorporate physiological measures and assess real-world skill transfer
Deep Learning Factor Investing in the Indonesian Stock Market
Full text linkTraditional linear factor models often fail to capture the complex, non-linear dynamics of emerging stock markets. This research designs and validates a novel Recurrence Plot (RP) matrices with β-VAE deep learning methodology to discover non-linear investment factors within the Indonesian context. We demonstrate that this framework is a systematically superior "factor factory" compared to a linear RP with PCA baseline, discovering twice as many high-quality factors (Sharpe > 0.3) and generating 7-fold more alpha on average. A key finding is the model's ability to disentangle high-frequency predictive signals (identified by SHAP) from more valuable, low-frequency profitable trends (validated by backtesting). The champion factor from this process yields a robust annualized alpha of 6.65% with a minimal max drawdown of -7.73% from 2018 to 2025. This study concludes that the RP -> β -VAE approach is a robust and resilient framework for discovering safer, non-linear sources of return unexplained by conventional models
Prostate Cancer Detection Using Gradient Boosting Machines Effectively
Full text linkProstate cancer remains a leading cause of cancer-related deaths among men globally, emphasizing the critical need for accurate diagnostic tools. This study investigates the application of Gradient Boosting Machines (GBMs) for prostate cancer detection using a dataset with key tumor characteristics such as radius, texture, area, and symmetry. Data preprocessing included normalization, missing value handling, and the Synthetic Minority Oversampling Technique (SMOTE) to address class imbalance. The GBM model demonstrated an accuracy of 75%, with high precision (82%) and recall (88%) for malignant cases, underscoring its potential as a reliable diagnostic tool. However, the model's performance for benign cases was limited by severe class imbalance, reflected in a precision of 33% and recall of 25%. Interpretability was enhanced using SHAP values, identifying key predictors like tumor perimeter and compactness. While GBMs show promise in prostate cancer diagnostics, future research should incorporate multimodal data, advanced balancing techniques, and rigorous validation frameworks to enhance generalizability and fairness. This study highlights the value of machine learning in healthcare, contributing to improved diagnostic accuracy and patient outcomes
Classifying Heart Disease through Fusion of Multi-Source Datasets: Integration of Feature Selection and Explainable Machine Learning Techniques
Full text linkThis study delves into heart disease classification through integrated feature selection and machine learning methodologies, utilizing three datasets comprising 4,728 participants and 11 features, with 4.27% missing data. Employing machine learning, we used XGBoost to achieve 0.95 accuracy for one feature, while Random Forest (RF) demonstrated accuracies of 0.92 and 0.99 for the remaining two features. Comparing 11 classification models, RF and XGBoost classified heart disease with 0.97 and 0.99 accuracy, respectively, using all available features. Applying Feature Elimination with Simultaneous Perturbation Feature Selection and Ranking (SpFSR) revealed that RF attained 0.99 accuracy by selecting only four features (cholesterol level, age, resting electrocardiographic measurements, and maximum heart rate), while XGBoost dropped to 0.91. Constructing an RF model with four features enhanced interpretability without compromising accuracy. Explainable Machine Learning (XAI) techniques, including Permutation Importance and SHAP Summary Plot analyses, gauged feature impact on heart disease prediction. The resting electrocardiographic measurements feature held the highest value (0.40 ± 0.01), followed by maximum heart rate (0.32 ± 0.01), cholesterol level (0.28 ± 0.01), and age (0.26 ± 0.005). These results underscore the significance of each feature in diagnosing heart disease via machine learning
DEVELOPMENT OF CHATBOT FOR PRE-DIAGNOSIS AND RECOMMENDATION OF ANXIETY DISORDER USING DIET AND SENTENCE TRANSFORMER MODELS
Full text link Previous research on chatbots for pre-diagnosis and recommendation of anxiety disorders has been limited to therapy aids. Comparing NLU DIET and LogisticRegressionClassifier models, this chatbot system calculates anxiety levels using GAD-7, DASS, and STAIT/STAIS-5 methods along with Sentence Transformer (SBERT) for semantic similarity.Intent classification testing yielded 95% accuracy for NLU DIETClassifier and 99% for LogisticRegressionClassifier. The Dialog Model achieved 68% accuracy with TEDPolicy. Testing involved 35 randomly selected respondents, including students and workers. From their interactions, the SBERT recommendation model scored 30% MAP, 26% for the Indobert base and paraphrase-multilingual-MiniLM-L12-v2 models.The average satisfaction and performance rating for the chatbot system was 3.7 out of 5. This research addresses the need for a prototype chatbot for pre-diagnosis and recommendation of anxiety disorders, with the best NLU model being LogisticRegressionClassifier at 99% accuracy and the dialog model at 68%. However, the recommendation system still has a low MAP due to the use of non-valid clinical data as references, suggesting room for improvement in future research
Obstacles Detection in Underwater Environment Using ROV Based on Convolutional Neural Network
Full text linkPada saat RoV berada dibawah air tidak sedikit obstacle yang dijumpai dan berpengaruh terhadap kinerja dan keselamatan body ROV itu sendiri. Obyek yang tertangkap kamera ROV seringkali sulit untuk diidentifikasi dan dideteksi karena besarnya noise bawah air. Selain itu, sifat air yang membiaskan cahaya dan tingkat kejernihan air turut berpengaruh terhadap kualitas gambar yang dihasilkan. Untuk membantu dalam mengidentifikasi obyek yang ada di bawah air, maka pada penelitian ini proses identifikasi dilakukan dengan menggunakan Convolutional Neural Networks (CNN). CNN mengekstraksi fitur penting dari gambar melalui beberapa lapisan konvolusi. Setiap lapisan konvolusi menggunakan filter untuk mendeteksi pola seperti tepi, sudut, atau tekstur dari gambar input. Pada tahap akhir, fitur-fitur yang sudah diproses ini dihubungkan ke lapisan fully-connected yang bertindak sebagai pengklasifikasi. CNN kemudian memetakan fitur-fitur tersebut ke dalam kelas-kelas tertentu , misalnya objek seperti botol, tiang kayu, rantai, dan propeller. Dari pengujian secara real-time sistem berhasil menunjukkan performansi yang baik dengan akurasi validasi sebesar 99.25% dan akurasi klasifikasi real-time sebesar 85%. Hasil klasifikasi selanjutnya menentukan pergerakan thruster ROV
Preprocessing Algorithm for K-Means Anomaly Detection on Payment Logs
Full text linkThe payment aggregator system with the single settlement feature enhances transaction efficiency. However, this also poses risks of cyberattacks and system errors. These risks can lead to abnormal events or anomalies. The middleware service records transaction activities in the form of logs. Log data can be analyzed for anomaly detection resulting from cyberattacks or system errors.K-Means clustering is less effective in detecting anomalies in log data because transaction log data is often unstructured, inconsistent, and has varying feature scales.This study develops a preprocessing algorithm to improve data quality before clustering. Transaction log data from July to December 2023 is used, with preprocessing stages including normalization, standardization, and Principal Component Analysis (PCA). K-Means is applied with K-Means++ initialization, and the number of clusters is determined using the kneedle algorithm. The results show that standardization improves segmentation, and PCA enhances anomaly detection effectiveness
Implementation of Chi-Square Feature Selection for Parkinson’s Disease Classification Using LightGBM
Full text linkPenyakit Parkinson merupakan penyakit yang disebabkan oleh kerusakan sel saraf otak dan termasuk penyakit yang jumlah kasusnya meningkat pesat di dunia. Salah satu cara yang dapat dilakukan untuk mencegah meningkatnya kasus penyakit Parkinson adalah dengan melakukan diagnosis melalui metode klasifikasi dengan pendekatan pembelajaran algoritmik. Penelitian ini mengimplementasikan teknik Chi-Square untuk pendekatan pemilihan fitur yang relevan dengan algoritma Light Gradient Boosting Machine (LightGBM) dalam klasifikasi penyakit Parkinson. Pemilihan fitur Chi-Square bertujuan untuk mengurangi fitur yang kurang relevan sehingga dapat meningkatkan hasil kinerja model. Selain itu, metode SMOTE diterapkan untuk menangani ketidakseimbangan data dan penyetelan hiperparameter guna menentukan kombinasi parameter yang optimal. Pengujian dilakukan terhadap sepuluh variasi jumlah fitur, dengan hasil terbaik diperoleh dengan menggunakan 200 fitur yang menghasilkan akurasi sebesar 96,05%. Dengan menggunakan metode Chi-Square, kinerja model LightGBM meningkat dibandingkan dengan kinerja tanpa pemilihan fitur. Penerapan kombinasi metode ini dapat meningkatkan kinerja model klasifikasi secara signifikan dan berpotensi untuk diterapkan dalam sistem pendukung diagnosis penyakit Parkinson