Publication Search

Background: The development of industrial technology in the Industry 4.0 era has encouraged the implementation of intelligent monitoring systems to improve machine reliability and operational efficiency. However, machine fault diagnosis systems based on artificial intelligence often face limitations in terms of interpretability because the models used are complex and difficult to explain. Objective: This study aims to develop a deep learning-based industrial machine fault diagnosis system integrated with an Explainable Artificial Intelligence (XAI) approach to improve diagnostic accuracy while providing interpretable insights for users. Method: The research method involves collecting data from industrial machine sensors consisting of vibration signals, temperature measurements, and acoustic signals, followed by data preprocessing and feature extraction processes. The processed data are then used to train a deep learning-based diagnostic model, after which explainability methods such as SHAP or LIME are applied to analyze the contribution of each feature to the model’s prediction results. Model performance is evaluated using accuracy, precision, recall, and F1-score metrics. Results: The results indicate that the proposed deep learning model achieves better performance compared to conventional machine learning methods such as Support Vector Machine and Random Forest. Furthermore, the explainability analysis reveals that vibration amplitude, increases in machine component temperature, and anomalies in acoustic signals are the main factors influencing machine fault detection. Therefore, the proposed system not only improves the accuracy of machine fault diagnosis but also provides transparency in the decision-making process, thereby supporting the implementation of predictive maintenance in smart manufacturing environments.

Communication is a fundamental human need for conveying information and ideas. However, individuals who are deaf and mute face difficulties in communicating with the broader community that does not understand sign language. This study aims to design and implement a real-time static sign language translator into speech using five flex sensors, an MPU6050 sensor, a Raspberry Pi Pico, an ADS1115 ADC module, and a DFPlayer Mini module as the audio output medium. Testing results show that the device successfully recognizes finger movements and hand orientation. The system is capable of playing audio output corresponding to recognized gestures, with the shortest latency recorded at 1.1 seconds and the longest at 2.8 seconds, achieving a detection accuracy rate of 75% based on 60 tests across 12 sign words. This device supports the translation of 12 simple sign words. The implementation demonstrates potential as an assistive communication tool, although further development is needed to improve accuracy, expand vocabulary, and conduct trials directly with deaf or mute users.

Background: The increasing complexity of industrial production systems requires machine condition monitoring solutions that are capable of operating in real time with high accuracy and responsiveness to support predictive maintenance strategies. Conventional cloud based monitoring systems often experience limitations such as high latency and dependence on stable network connectivity, which can delay decision making processes in critical industrial operations. Objective: This study aims to design and evaluate an Industrial Internet of Things (IIoT) architecture based on edge computing to improve the efficiency of industrial sensor data processing and accelerate anomaly detection in industrial machines. Method: The research adopts an experimental approach by designing a system architecture consisting of a sensor layer, edge computing layer, and cloud layer. Industrial sensors, including vibration, temperature, and current sensors, continuously collect machine operational data, which are then processed locally at the edge node using a machine learning based anomaly detection algorithm. System testing is conducted in a simulated manufacturing environment to evaluate performance based on latency, reliability, and detection accuracy. Results: The results indicate that edge based data processing significantly reduces latency compared with cloud-based processing and enables faster responses to machine condition changes. Additionally, the implemented anomaly detection algorithm achieves high accuracy in identifying abnormal sensor data patterns.

Hypertension is a major global health risk that requires accurate early detection, yet conventional methods struggle with complex and imbalanced health datasets. This study aims to optimize hypertension prediction using a Logistic Regression model integrated with Borderline-SMOTE to enhance recall and provide model transparency through SHAP (Shapley Additive Explanations). The method utilizes the BRFSS dataset, applying Borderline-SMOTE to address class imbalance at the decision boundary and XAI techniques for global and local interpretation. The findings show that the model achieved an accuracy of 0.719, an AUC of 0.800, and a significantly improved recall of 0.756. SHAP analysis identified age, high cholesterol, and BMI as the most influential risk factors, while waterfall plots successfully clarified individual risk extremes, ranging from 1.72% to 99.43% probability. These results imply that the proposed approach provides a sensitive and transparent screening tool for public health practitioners, effectively balancing statistical efficiency with clinical accountability.

The increasing volume of online hotel reviews provides valuable insights into customer perceptions but poses challenges for manual analysis due to its unstructured nature. This study aims to compare the performance of Recurrent Neural Network (RNN) and Bidirectional Encoder Representations from Transformers (BERT) in hotel review sentiment analysis. A total of 20,491 TripAdvisor hotel reviews were classified into three sentiment categories: negative, neutral, and positive. The research methodology includes text preprocessing, stratified data splitting, class imbalance handling using Random Over-Sampling, tokenization, and supervised model training. Model performance was evaluated using a confusion matrix and classification metrics. The results indicate that BERT outperforms RNN, achieving an accuracy of 80.54%, while RNN reached 62.21%. BERT demonstrated superior capability in capturing contextual and semantic information in hotel reviews. These findings suggest that transformer-based models are more effective for sentiment analysis of complex textual data in the hospitality domain and can support data-driven service improvement strategies.    

This study develops an interpretable machine learning model to predict the settlement status of Hajj fees in Jambi Province, Indonesia. Utilizing the XGBoost algorithm on a dataset of 4,332 prospective pilgrims from 2025, the research addresses the critical challenge of class imbalance where only 28.5% of samples are labeled "Unsettled". The baseline XGBoost model achieved a ROC-AUC of 0.7778, with a recall of 0.3482 for the minority class. SHAP (SHapley Additive exPlanations) analysis was employed to interpret model predictions, revealing that financial features specifically NILAI_VA (Virtual Account Value), JML_SETORAN (Deposit Amount), and JML_PELUNASAN (Settlement Amount) are the most significant factors influencing repayment risk, with negative SHAP values indicating increased default probability. The findings demonstrate that an interpretable XGBoost framework can provide both predictive accuracy and actionable insights for policymakers, enabling targeted interventions such as flexible payment schemes and enhanced financial monitoring for high-risk pilgrims..

WPS Office stands for Writer, Presentation, and Spreadsheets—a software suite offering diverse office functions, including document processing, spreadsheet creation, and presentation tools. This study analyzes user satisfaction levels and the influence of the variables Content, Accuracy, Format, Ease of Use, and Timeliness on WPS Office application users in Jambi City, using the End User Computing Satisfaction method. Data were gathered through an online questionnaire distributed to students in Jambi City who had used the application; created via Google Forms, it garnered 385 responses. Post-collection, analysis was conducted using Structural Equation Modeling in SmartPLS software version 4. Of the five hypotheses tested, four were accepted. The results reveal that accuracy, format, ease of use, and timeliness positively and significantly influence user satisfaction, while content shows no significant effect.

The Indonesia Pintar Program (PIP) is an educational assistance program for students from underprivileged families, but determining the eligibility of recipients still faces obstacles in the form of subjectivity and data imbalance. This study aims to classify the eligibility of high school students receiving PIP in Jambi City using data mining methods. The SMOTE technique was applied to overcome class imbalance, and Gain Ratio feature selection was used to determine important attributes. The dataset used consisted of 19,596 student data with a training data distribution of 70% and testing data of 30%. The classification process used the Naïve Bayes, Decision Tree (J48), and Random Forest algorithms with the Use Training Set, 5-Fold, and 10-Fold Cross Validation testing schemes. The results show that SMOTE improves model performance, but feature selection in some cases reduces accuracy. Overall, Random Forest without feature selection provides the best results with an accuracy of 93.33% and is recommended as the most effective model for objectively determining PIP recipient eligibility.

Sentiment analysis is a technique in Natural Language Processing (NLP) used to identify user opinion tendencies based on textual reviews. This study analyzer user reviews of the Maxim application on the Google Play Store and compares three Machine Learning algoritmhs-Naïve Bayes, Support Vector Machine (SVM), and CatBoost-in classifying sentiment. The research stages include data collection, text preprocessing, feature extraction using TF-IDF and Chi-Square, class balancing using SMOTE, and performance evaluation through Accuracy, Precision, Recall, and F1-Score. ANOVA is used to examine the influence of feature selection on model performance. The results show that each model exhibits different performance level across the tested feature combinations. The CatBoost achieved the highest accuracy of 99,26% and demonstrating the most stable performance. Meanwhile, the Naïve Bayes and SVM models experienced performance decreases experiments, especially after applying SMOTE. These findings indicate that the choise of algorithm, feature extraction method, and class balancing technique significantly affects classification outcomes. Overall, CatBoost is identified as the best-performing model, providing more consistenst classification result in accordance with the characteristics of the user reviews.

The development of digital technology has led to an increase in the use of short video-based entertainment applications, including the Melolo application. However, the free version still has various complaints, such as inconsistent subtitles, unintuitive navigation, force close glitches, and unstable advertisements, so user satisfaction analysis is needed. This study aims to measure the level of satisfaction of users of the free version of the Melolo application using the End User Computing Satisfaction (EUCS) method, which covers five variables, namely content, accuracy, format, ease of use, and timeliness. Data was collected through an online questionnaire of 385 Melolo app users in Jambi City and analyzed using Structural Equation Modeling (SEM) with the help of SmartPLS 4. The results showed an R-Square value of 0.546, indicating that the model was able to explain 54.6% of the changes in user satisfaction levels. The variables of content and timeliness were found to have a significant effect on user satisfaction, while accuracy, format, and ease of use had no significant effect. These results indicate that content quality and system timeliness are the main factors in increasing user satisfaction. Therefore, Melolo app developers are advised to maintain content quality and improve system performance and stability to optimize the user experience.

The implementation of the One-Stop Integrated Service (PTSP) Website at the Ministry of Religious Affairs of Jambi City is part of the digital transformation of public services. This study aims to measure user satisfaction with the PTSP Website using the End User Computing Satisfaction (EUCS) method and the DeLone & McLean model. This research employed a quantitative approach using a survey method involving 100 respondents who are users of the PTSP Website. The EUCS variables consist of content, accuracy, format, ease of use, and timeliness, while the DeLone & McLean model includes system quality, information quality, and service quality. Data were analyzed using Partial Least Squares–Structural Equation Modeling (PLS-SEM) with SmartPLS 4.0 software. The results indicate that system quality, information quality, and service quality have a positive effect on user satisfaction with the PTSP Website. This study is expected to serve as an evaluation reference for improving the quality of digital public services.

Student selection for the National Student Sports Olympiad (O2SN), particularly in the karate category, is still largely conducted manually, which may lead to subjectivity and inconsistency in the assessment process. This study aims to design and develop a web-based Decision Support System (DSS) to assist schools in selecting the best students for the O2SN karate competition by applying the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method at SMA Negeri 2 Muaro Jambi. The system was developed using the waterfall model, which consists of requirement analysis, system design, implementation, and testing stages. The student evaluation process is based on criteria referring to the official O2SN karate standards, including stance accuracy, techniques, movement transitions, timing and harmony, breathing control, focus (kime), conformity (consistency with ryu-ha/style), strength, speed, and balance. The developed system processes assessment data to generate preference values and student rankings, which are separated by male and female categories. The results indicate that the application of the TOPSIS method is able to support more objective decision-making and improve transparency and efficiency in the O2SN student selection process within the school environment.

Student dropout remains a significant challenge for higher education institutions as it impacts academic quality, educational management efficiency, and students' success in completing their studies. Therefore, an approach that can identify students at risk of dropping out is necessary so that timely academic interventions can be made. This study aims to develop a dropout detection model using an Artificial Neural Network (ANN). The data used come from a publicly available higher education dataset, ensuring research reproducibility. Data preprocessing steps were carried out to improve data quality before modeling, and the Synthetic Minority Over-Sampling Technique combined with Edited Nearest Neighbors (SMOTE-ENN) was applied to address class imbalance issues. The ANN model's performance was evaluated using accuracy, precision, recall, F1-score, and area under the ROC curve (ROC-AUC). The test results show that the ANN model can provide excellent predictive performance in detecting at-risk students. The application of SMOTE-ENN also proved to enhance the model’s sensitivity toward the minority class, as indicated by improvements in recall and F1-score. These findings indicate that the developed ANN model has the potential to be used as a student dropout detection system to support data-driven decision-making and strategy development within higher education institutions.

The rapid growth of e-commerce platforms in Indonesia, particularly Tokopedia, has resulted in a large volume of consumer reviews containing valuable information regarding customer perceptions and satisfaction. However, manual analysis of such reviews is inefficient and prone to subjectivity, necessitating an automated approach based on machine learning. This study aims to classify the sentiment of sports product reviews on Tokopedia into positive, negative, and neutral categories by applying Logistic Regression, Support Vector Machine (SVM), and Random Forest using the Term Frequency–Inverse Document Frequency (TF-IDF) approach. The data were collected through web scraping of Indonesian-language sports product reviews and processed through several preprocessing stages, including data cleaning, case folding, tokenization, stopword removal, and stemming. Feature representation was performed using TF-IDF to transform textual data into numerical vectors, after which the dataset was divided into training and testing sets with an 80:20 ratio. Model performance was evaluated using accuracy, precision, recall, and F1-score metrics. The results indicate that the application of TF-IDF significantly improves the performance of all models, with SVM consistently achieving the most optimal performance compared to Logistic Regression and Random Forest. These findings demonstrate that classical machine learning algorithms combined with TF-IDF remain highly effective for sentiment analysis of Indonesian-language text. The implications of this study are expected to assist sellers in understanding customer opinions, support consumers in making informed purchasing decisions, and serve as a foundation for the development of sentiment analysis and recommendation systems on e-commerce platforms.

This study analyzes creditworthiness assessment and predicts non-performing loan (NPL) risk using the Naïve Bayes algorithm at BPR Ukabima Lestari, Jambi Branch. A quantitative data mining approach with probabilistic classification is applied. The dataset includes borrower attributes such as age, occupation, income, loan amount, tenor, collateral, and repayment history. Research stages comprise data preprocessing, model development, and performance evaluation using accuracy, precision, recall, and F1-score implemented in RapidMiner. The results indicate that the Naïve Bayes model achieves 99.58% accuracy, demonstrating strong capability to predict potential problem loans accurately and efficiently, supporting data-driven credit decisions and strengthening credit risk management in microbanking institutions.

The growth of e-commerce in Indonesia has led to an increase in product reviews, including for beauty products on Tokopedia and Shopee. These reviews serve as important sources of information to assess consumer satisfaction; however, manually analyzing thousands of reviews daily is impractical. This study applies Natural Language Processing (NLP) with Naive Bayes, C4.5, XGBoost algorithms to classify sentiment in Indonesian-language reviews. The dataset used consists of 76,256 reviews labeled as positive, negative, and neutral. The research stages include text preprocessing, feature representation using BoW and TF-IDF, data balancing through SMOTE, and model performance evaluation based on accuracy, precision, and recall. Differences in results among the algorithms were analyzed using ANOVA. The results show that Naive Bayes achieved the highest accuracy at 67.71%, followed by XGBoost at 65.91%, and C4.5 at 58.39%, with Naive Bayes performing best in identifying positive and negative sentiments, while XGBoost and C4.5 handled more complex data patterns effectively. These findings provide guidance for sentiment analysis in Indonesian and support businesses in obtaining automated insights from customer reviews to improve product quality and services.

This study conducts sentiment analysis of Indonesian user reviews of the CapCut application using IndoBERT and compares two evaluation schemes: a single 80/20 train–test split and stratified 5-fold cross-validation (k=5). A total of 1,048,575 reviews were collected from the Google Play Store through web scraping and labeled into three sentiment classes based on rating: negative (1–2), neutral (3), and positive (4–5). After preprocessing—cleaning, case folding, banned-word removal, normalization—and duplicate removal, 517,962 reviews were retained. IndoBERT Base P1 was fine-tuned using fixed hyperparameters (batch size 32, learning rate 2e-5, up to 4 epochs, early stopping patience 2), while undersampling was applied to the training set to address class imbalance. Performance was assessed using accuracy, precision, recall, F1-score, and ROC-AUC, supported by confusion matrix and ROC-curve visualizations. The single split achieved an accuracy of 0.756, whereas cross-validation produced a mean accuracy of 0.740. Across both schemes, the positive class achieved the best performance (F1-score 0.850; ROC-AUC 0.918–0.919), while the neutral class remained the most challenging (precision 0.198–0.206; F1-score 0.280–0.283). Overall, cross-validation is recommended for reporting because it reduces dependence on a single partition and provides a more representative estimate across multiple splits.

Jambi Province is one of the largest natural rubber producing regions in Indonesia; however, rubber factories under GAPKINDO Jambi still face productivity issues, particularly the gap between production capacity and actual output, and productivity assessment that is still conducted manually by GAPKINDO Jambi. This study employs Decision Tree, Random Forest, KNN, and SVM algorithms within a structured pipeline involving preprocessing, feature selection, standardization, data balancing using SMOTE, and hyperparameter tuning. The proposed solution applies productivity level classification both individually and through paired combinations (ensemble voting). The results show that the Decision Tree + Random Forest model achieves the best performance with an accuracy of 0.84 and an F1-score of 0.83, confirming the effectiveness of ensemble methods in supporting productivity improvement decisions.

The massive adoption of Internet of Things (IoT) devices is expanding the cyberattacks surface, particularly by the Mirai botnet, which exploits the dynamic characteristics of data traffic. This research proposes a Mirai detection approach based on a Recurrent Neural Network (RNN) optimized using Bayesian Optimization to improve prediction accuracy on sequential data. Unlike previous studies, this research utilizes the latest CIC IoT-DIAD 2024 dataset and applies probabilistic optimization to the hyperparameter space, including RNN units, dropout, and learning rate. The experiment was conducted on 201,021 valid data points, with dimensionality reduction using PCA as the optimal point to represent essential features without redundancy. The results show a significant increase in accuracy from 97.95% to 99.69%, accompanied by an 84% decrease in False Negatives, an 86% decrease in False Positives, and an AUC value of 0.9999. These findings confirm that integrating RNN and Bayesian Optimization not only improves numerical performance but also strengthens the reliability of the intrusion detection system for modern IoT ecosystems with controlled computational loads.

This study aims to classify customer payment methods at 17 Coffee & Eatery using machine learning algorithms, namely Naïve Bayes and Support Vector Machine (SVM). The increasing use of digital and non-cash payments has generated large volumes of transaction data that are rarely analyzed optimally, even though such data contain valuable information for business decision making. This research used secondary transaction data collected from January to March 2025, consisting of 10,147 transaction records. The dataset included several attributes such as order time, payment time, transaction type, total sales, number of items, and payment method. Data preprocessing was performed through data cleaning, feature engineering, normalization, and label encoding before being divided into training and testing sets with an 80:20 ratio. The Naïve Bayes and SVM models were then trained and evaluated using accuracy, precision, recall, F1-score, and ROC–AUC metrics. The results show that both algorithms were able to classify payment methods effectively, but SVM achieved higher accuracy and more stable performance than Naïve Bayes. These findings indicate that SVM is more suitable for handling complex and heterogeneous transaction patterns. The implementation of machine learning for transaction classification can support more efficient financial management and data-driven decision making for small and medium enterprises in the culinary sector.