Publication Search

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..

Gestational Diabetes Mellitus (GDM) is a pregnancy-related metabolic disorder that poses health risks to both mother and fetus if not detected early, requiring accurate prediction methods for early screening and clinical decision-making. This study applies the Random Forest algorithm to detect GDM risk using clinical data from the Pima Indian Dataset. Data preprocessing included handling missing values, standardization, feature engineering, and a 70:30 train–test split. Two models were developed: a baseline and an optimized model using GridSearchCV hyperparameter tuning, validated with 5-fold cross-validation. Performance was assessed using a classification report, confusion matrix, and ROC–AUC. Results show that the optimized model outperforms the baseline, achieving 88% accuracy, an AUC of  93%, and average recall of 81%–85%. Compared to previous studies, this approach demonstrates improved predictive performance. The findings indicate that combining Random Forest with comprehensive preprocessing, feature engineering, and model optimization is effective and feasible for developing a medical decision support system for early GDM risk screening.

Diseases in primary health services exhibit complex spatial-temporal dynamics due to urbanization and population mobility. Conventional surveillance approaches are difficult to capture these patterns adaptively. Machine learning (ML) based on spatio-temporal modeling offers a solution with the ability to detect disease clusters automatically and with high precision. Research Objectives: This research aims to develop a machine learning model to detect disease hotspots from primary service data in Indonesia, with a focus on improving prediction accuracy, interpretability, and relevance of health policies. Methodology: The primary service dataset for 2024 (5,343 entries) was analyzed using three ML models Gradient Boosting Machine (GBM), Temporal Random Forest (TRF), and Multi-EigenSpot with spatial (village) and temporal (week, month) features. Performance evaluation includes predictive (AUC, F1-score) and spatial (Moran's I, Spatio-Temporal Correlation Index) metrics. Results: The results showed that Multi-EigenSpot achieved the best performance (AUC=0.91; F1=0.86), with the detection of dominant hotspots in Sungai Asam and Beringin Villages. Moran's I value of 0.63 indicates a strong spatial autocorrelation, while STCI=0.57 indicates moderate temporal stability. Conclusions: ML-based spatio-temporal models are effective in identifying hidden disease patterns and have the potential to be integrated into national digital surveillance systems. This approach supports precision public health by providing a scientific basis for real-time location- and time-based intervention policies.

End-to-end autonomous driving has emerged as a promising paradigm in which deep neural networks directly map raw visual inputs to continuous control actions. Despite its effectiveness, this approach suffers from limited transparency, posing significant challenges for deployment in safety-critical driving scenarios. This study addresses the lack of interpretability in vision-based end-to-end autonomous driving systems and aims to analyze model decision-making behavior under critical conditions such as sharp steering maneuvers and abrupt control transitions. To this end, an explainable end-to-end autonomous driving framework is proposed, combining a convolutional neural network trained via imitation learning with gradient-based visual attribution techniques, including Grad-CAM. The model predicts continuous steering, throttle, and braking commands directly from front-facing camera images, while explainability mechanisms are applied to reveal input regions influencing each control decision. Model performance is evaluated using both prediction accuracy and safety-oriented behavioral metrics. Experimental results show that the proposed explainable model achieves lower control prediction errors compared to a baseline end-to-end CNN, reducing steering mean squared error from 0.034 to 0.031, throttle error from 0.021 to 0.019, and brake error from 0.018 to 0.016. Moreover, safety-oriented analysis indicates improved driving stability, with steering variance reduced from 0.087 to 0.072 and abrupt control changes decreased from 14.6 to 10.3 events. Visual explanations consistently highlight road surfaces and lane-related structures during complex maneuvers, indicating reliance on semantically meaningful cues. In conclusion, the results demonstrate that integrating explainability into end-to-end autonomous driving not only preserves predictive performance but also correlates with smoother and more stable driving behavior. This framework contributes to the development of transparent and trustworthy autonomous driving systems suitable for safety-critical applications

This research investigates the Peak Particle Velocity (PPV) resulting from blasting operations at Pit Pinang, PT Bukit Baiduri Energi, employing two prediction approaches: Non-Linear Geometric Regression and the USBM Oriard’s Formula. Ground vibration measurements were recorded over a one-month period, from October 9 to November 8, 2025. The findings indicate that the non-linear regression model achieves a higher predictive accuracy of 78.62%, outperforming the USBM Oriard’s Formula, which reaches 68.2%. Variations between the observed and estimated PPV values were affected by factors such as the location of geophones, differences in explosive charges, and alterations in borehole depths. In addition, the study evaluates optimal explosive charge recommendations in accordance with SNI 7571:2010 standards to mitigate potential structural damage in surrounding areas. By highlighting these predictive discrepancies and providing practical guidance on charge management, the research offers valuable insights for improving blasting safety and minimizing vibration impacts on nearby infrastructure. The comparison of methods emphasizes the importance of selecting appropriate prediction models to ensure both operational efficiency and environmental safety.

The disparity in the quality of rehabilitation services across regional work units presents a significant challenge to effective public management. This study aims to bridge the gap between problem diagnosis and policy prediction by proposing a hybrid, data-driven approach. We integrate K-Means Clustering to map the current state of service quality and Stochastic Simulation to predict the impact of strategic interventions. Using the 2024 Public Satisfaction Index (IKM) dataset from the National Narcotics Agency (BNN), the K-Means algorithm initially identified 26 work units (15.7%) in the "Red Zone" (critical performance), highlighting urgent areas for improvement. Next, a stochastic simulation modeling a "Directed Priority Intervention" scenario was run. The results predicted a significant structural shift in the distribution of service quality, characterized by an 80.8% decrease in critical units (down to 5 units) and a 71.8% increase in excellent performing units (up to 67 units). These findings validate that the integration of clustering and simulation provides a comprehensive framework for evidence-based decision-making, enabling policymakers to optimize resource allocation and efficiently accelerate national service standardization.

This study aims to estimate changes in the Indonesian Sharia Stock Index (ISSI) from 2020 to 2025 using the Autoregressive Integrated Moving Average (ARIMA) model. The growth of the Islamic stock market in Indonesia has increased rapidly, driven by public awareness of investments that follow sharia principles, as well as changes in macro and microeconomic conditions, especially during the COVID-19 pandemic which has had a significant impact on the financial market. This study relies on monthly ISSI data taken from official sources and analyzed with a quantitative approach using the time series method using EViews version 13 software. Statistical analysis and stationarity tests indicate that the ISSI data exhibits an increasing trend pattern and quite high volatility, so that a differentiation process is necessary to achieve stationarity. Based on the results of model testing and the selection of optimal information criteria, the ARIMA (1,1,1) model was selected as the most appropriate to capture the autocorrelation pattern and produce accurate short-term predictions. Projections indicate a stable growth trend until the end of 2025, with an estimated index of more than 8.3 million. The findings of this study indicate that the ARIMA model is an effective tool for forecasting ISSI movements and can be a strategic consideration for investors, financial institutions, and policymakers in developing sustainable investment strategies in the Indonesian Islamic stock market.

Penelitian ini menggunakan data sekunder yang telah melalui beberapa proses pra-pengolahan, mencakup penanganan data yang hilang, standarisasi data numerik, serta konversi data kategorikal menggunakan teknik One-Hot Encoding. Sebagian besar data (80%) digunakan dalam tahap pelatihan, sedangkan 20% sisanya digunakan untuk tahap pengujian, sedangkan model diimplementasikan dengan metode LinearRegression() pada library scikit-learn. Hasil evaluasi menunjukkan bahwa model berhasil menangkap hubungan linier di antara variabel independen dan dependen, memperoleh nilai MAE = 0,509; MSE = 0,464; RMSE = 0,681; dan R² = 0,627. Hal ini menandakan bahwa sekitar 62,7 persen variasi harga rumah di wilayah Jabodetabek dapat dijelaskan oleh model tersebut.

Rubber prices experience significant and prolonged fluctuations, which impact farmer incomes and management decisions. Understanding historical patterns and price predictions is considered crucial for production planning, marketing, and farmer protection policies. This study aims to identify the characteristics of rubber price time series in Lahat Regency and develop a reliable forecasting model to support short- to medium-term decision-making. This study uses secondary data on monthly average producer prices for the period January 2019–December 2023. The analysis includes the Augmented Dickey–Fuller stationarity test to determine the need for transformation, differencing, and/or logarithmic transformation when necessary, identification of autocorrelation patterns using ACF/PACF, model estimation on the processed data, and evaluation of residual diagnostics (Ljung–Box, normality test) and forecasting accuracy metrics (RMSE, MAE, MAPE, Theil). The level data shows non-stationarity and becomes stationary after the first differencing; The model on log-transformed data had significant parameters and higher explanatory power than the model on de-differenced data, with RMSE and MAPE values ​​within a reasonable range. Forecast confidence intervals widened at longer time horizons, indicating increased projection uncertainty. Conclusion: Validated forecasts can inform farmers and policymakers to manage price risk and design market interventions.

Production quantity planning is a crucial component in the bottled water industry (AMDK) to ensure that consumer demand is met efficiently. Inaccuracies in determining the amount of production can lead to overproduction and supply shortages, which ultimately leads to increased operational costs and decreased customer satisfaction. This study applies the Sugeno fuzzy logic method to predict the amount of production based on two main variables, namely weekly demand and raw material stock. The analysis stages include the fuzzification process, the preparation of the rule base, inference using the zero-order Sugeno method, and defuzzification using the Weighted Average (WA) method. The data used is synthetic data that represents the operational conditions of the medium-scale bottled water industry. The results show that the Sugeno fuzzy system is able to produce production predictions that are adaptive and responsive to fluctuations in demand and stock availability. This model provides consistent and stable output, so it can help companies in determining the optimal production amount. These findings confirm that Sugino's fuzzy approach can be an effective decision support tool in bottled water production management, especially in the face of uncertainty and variability in market demand.

Chronic kidney disease is an increasingly prevalent health issue that requires more precise clinical data-based early detection methods to enable timely and appropriate treatment. This study focuses on developing a predictive model for chronic kidney disease using the Light Gradient Boosting Machine (LightGBM) algorithm and enhancing its performance through hyperparameter optimization with the Grey Wolf Optimizer (GWO). The dataset used originates from public sources and undergoes several preprocessing steps, including missing value imputation, categorical feature encoding, outlier handling, initial feature selection, and stratified data splitting to maintain model quality. Three modeling approaches were evaluated: LightGBM with default parameters, LightGBM enhanced using Random Search, and LightGBM optimized with GWO. The experimental results indicate that the baseline model already performs well, Random Search improves accuracy and F1-score, and GWO achieves the highest AUC-ROC value despite requiring longer computation time. Significance testing through cross-validation shows that the performance differences among the three models are not statistically significant, suggesting that the observed improvements are not strong enough to determine a definitively superior optimization method. The feature importance analysis highlights that clinical indicators such as creatinine levels, glomerular filtration rate, blood pressure, and urine protein contribute most prominently to the prediction. Overall, the study demonstrates that LightGBM is a reliable model for early detection of chronic kidney disease, and hyperparameter optimization still offers added value that can support the development of AI-based clinical decision-support systems

Accurate wind turbine power curve modeling plays a crucial role in performance evaluation, energy yield estimation, and data-driven control strategies. However, actual power curves often exhibit non-linear behavior influenced by atmospheric variability, measurement noise, and SCADA anomalies, making conventional modeling approaches less effective. This study proposes an optimized logistic power curve model whose parameters are tuned using Particle Swarm Optimization (PSO) to improve predictive accuracy. The analysis uses the Wind Turbine SCADA Dataset from Kaggle, which undergoes extensive preprocessing including physical rule filtering, outlier detection with the Interquartile Range (IQR) method, anomaly removal, and smoothing of the power signal. A three-parameter logistic model is selected due to its ability to capture the typical S-shaped relationship between wind speed and power output. PSO is applied to identify optimal model parameters by minimizing the Mean Squared Error (MSE), utilizing 40 particles over 200 iterations. The optimized model achieves strong predictive performance with RMSE of 404.09, MAE of 179.96, and R² of 0.904 on the test set, indicating that more than 90% of the variability in actual power can be explained by wind speed. Residual analysis reveals heteroscedastic patterns and slight overestimation in mid-range wind speeds, yet overall model consistency remains high. Comparative evaluation against Linear Regression, Random Forest, and logistic modeling using curve_fit shows that the Logistic–PSO approach provides the most accurate and stable predictions. These findings demonstrate that combining logistic modeling with PSO offers an effective and robust method for data-driven wind turbine power curve optimization.

Frequent damage to the grinding disc components and the transmission system of the 15 kg/hour rice flour grinder is a problem faced. The purpose of scheduling component maintenance and repairs is to obtain predictions of maintenance and repair schedules and costs for the period 2026. The component maintenance and repair scheduling method includes examining previous period maintenance and repair data, applying the inspection-replace-repair-overhaul (IRRO) method, assessing component conditions, estimating component life, estimating technician costs, estimating supporting work equipment and supporting materials to be used in maintenance, estimating the time for replacing spare parts or reinstalling components after repair, estimating maintenance and repair costs in 2026, and calculating the maintenance cost to profit ratio. The results of component maintenance and repair scheduling show that the maintenance cost in 2026 is Rp 1,370,000,- with an estimated annual profit potential of Rp 21,600,000, and the maintenance cost to profit ratio is 6.3%, which implies that the 15 kg/hour rice flour grinder is still quite prospective and feasible to use for the next few years.

This literature study evaluates the accuracy of the Slope Mass Rating (SMR) method for coal mine slope stability in Indonesia through a systematic descriptive synthesis of 25 empirical studies from 2020 to 2025. The objectives of the study were to identify the level of SMR prediction accuracy, factors affecting the method's performance, and modifications required for local Indonesian conditions. The research method involved a systematic search with inclusion criteria for empirical studies reporting SMR and/or Safety Factor (SF) values ​​for coal mines and associated slopes in Indonesia. Quantitative analysis showed a range of reported SMR values ​​between 41 and 96 with a median of 72, while SF values ​​ranged from 1.137 to 4.09 for normal operational conditions. The synthesis results indicated that SMR provides a consistent stability classification for initial slope design and failure mode identification (planar, wedge, toppling), with historical validation showing a correlation of up to 91.23% between SMR-based hazard zoning and actual field events in some cases. Key limitations include dependence on discontinuity data quality, sensitivity to groundwater conditions and tropical weathering, and variation in the interpretation of adjustment factors F1-F4. Modifications such as NAAF23 and integration with numerical modeling have been shown to improve prediction reliability. It is recommended that coal mining practitioners combine SMR with kinematic analysis and limit equilibrium modeling as standard operating procedures, and develop adjustment factors specific to Indonesian geological conditions. Further research should focus on standardizing parameter reporting and cross-site quantitative validation to enable more robust statistical meta-analyses.  

Mental health has become a major global issue, particularly after the COVID-19 pandemic, which significantly increased the prevalence of psychological disorders. Early detection of stress and other mental health problems remains a major challenge, as traditional methods are generally subjective and unable to provide real-time results. This study aims to design and test a wearable sensor based on Heart Rate Variability (HRV) as a physiological indicator for detecting stress levels. The research employed an experimental approach through the development of a wearable sensor prototype equipped with a stress detection algorithm based on HRV analysis, including both time-domain and frequency-domain parameters. The prototype was tested on 100 respondents with varying stress levels under controlled conditions. Instruments used in this study included the HRV sensor prototype, psychological questionnaires, and standard validation devices. Data were analyzed by comparing the sensor detection results with respondents’ psychological data and calculating prediction accuracy. The findings showed that the wearable sensor was able to predict stress conditions with an accuracy rate of 80%. The distribution of sensor detection results was generally consistent with psychological data, especially in the low-stress category, although slight deviations were observed in moderate and high-stress categories. These results demonstrate that an HRV-based wearable sensor can serve as a practical and non-invasive tool to monitor mental conditions in real time. The implications of this research highlight the potential of wearable technology as an innovative solution for mental health monitoring, both for individual use and as support for healthcare systems. Therefore, this study contributes to the development of adaptive and responsive health technologies in addressing global mental health challenges.

This study aims to implement an information retrieval system for cryptocurrency data using an attribute-based approach integrated with the Vector Space Model (VSM). The primary objective is to develop a system capable of retrieving the most relevant digital asset information according to specific search attributes, including positive sentiment, price fluctuation, and prediction confidence level. The research adopts a descriptive qualitative method combined with an experimental approach to evaluate the retrieval performance of the cosine similarity algorithm on normalized numerical data. Data preprocessing and attribute weighting were conducted to ensure consistency and improve retrieval accuracy. The experiment demonstrates that the proposed system achieves a Precision@5 value of 1.0, which indicates that all top-five retrieved results are fully relevant to user queries. These findings validate the effectiveness of the attribute-based VSM in analyzing multidimensional cryptocurrency datasets. Overall, this research contributes to the advancement of information retrieval applications in the cryptocurrency domain, particularly for supporting data-driven decision-making and intelligent financial analysis.

This classroom action research (CAR) aims to improve students’ English-speaking skills in formal communication contexts. The study focuses on second-semester students of the Business and Professional Communication Study Program at Politeknik Negeri Madiun in the 2024/2025 academic year. The main problem addressed is the students’ low ability to speak English formally and fluently. The objective is to determine whether watching YouTube videos can enhance students’ formal speaking performance. The research employed both qualitative and quantitative methods. Qualitative data were obtained from students’ performance scores before and during the implementation, while quantitative data were gathered through observations, questionnaires, and interviews to compare pre-test and post-test results. The intervention used several active viewing techniques, including active viewing, freeze framing and prediction, silent viewing, and reproduction activity. The findings indicate a consistent improvement in students’ speaking abilities across all research cycles. The average scores increased from 61.1 in the pre-test, to 63.6 in post-test Cycle I, 66.6 in Cycle II, and 70.0 in Cycle III. These results demonstrate that integrating YouTube videos with active viewing techniques effectively enhances students’ formal speaking competence. In conclusion, the use of interactive video-based learning can significantly support the development of English-speaking skills in formal contexts for vocational students.

An answer to the worldwide need for solutions to food security, data fusion technology that combines climate data with satellite imagery greatly improves the accuracy of agricultural yield predictions; this study intends to examine the advancements, methods, and key contributions of this area. By sifting through 62 papers pulled from Scopus, this research employs the SLR methodology. Document type, data source, open access, subject area, and year of publication (2020–2024) are some of the categories filtered through by Boolean keywords in the selection process. To assess patterns in publications, the efficacy of machine learning models, and key contributions, bibliometric analysis was performed. An upward tendency in publication has been identified by the analysis, particularly beyond the year 2023. Integrating geographical and temporal data has been a great success with machine learning models like Random Forest, Random Forest, and Gradient Boosting. Data resolution, integration of data from several sources, and a real-time framework are still missing pieces to the puzzle when it comes to generalizing research outcomes. More complex data fusion approaches, multiregional datasets, and advanced machine learning models to back more accurate agricultural predictions are all things that this study notes as needing additional investigation in the future. To further innovate agricultural yield prediction, multidisciplinary collaboration is also crucial.

This study aims to predict the total population of Deli Serdang Regency for the year 2025 using a multiple linear regression approach. The data used were obtained from the Central Bureau of Statistics (BPS) of Deli Serdang for the years 2015–2024, with total population as the dependent variable and male population and the percentage of male population as the independent variables. The analysis was carried out through a series of basic assumption tests, including normality, multicollinearity, heteroscedasticity, and autocorrelation, all of which indicated that the model met the criteria for a valid regression model. The results of the F-test and t-test showed that both independent variables had a significant influence on the total population. The R² value of 1.000 indicates that the model is capable of explaining 100% of the variation in the population size. Based on the regression model obtained, the projected total population of Deli Serdang in 2025 is estimated to reach 4,075,362 people, an increase of 2,026,882 people from the previous year. These findings are expected to serve as a basis for regional development planning, particularly in the provision of public services and resource management.

Academic planning is one form of planning the teaching and learning process in state universities, aimed at achieving educational goals based on the standards set. One important aspect of academic planning is forecasting the number of new students. This study compares two forecasting methods, Fuzzy Time Series (FTS) and Autoregressive Integrated Moving Average (ARIMA), in predicting the number of new students in the Statistics Study Program at Universitas Negeri Gorontalo. Forecasting the number of new students is crucial for determining various policies, such as resource allocation and providing adequate facilities. The results of the study show that the ARIMA method produces more accurate predictions with a Mean Absolute Percentage Error (MAPE) of 0.35%, which is lower than the FTS method. This indicates that ARIMA is more effective in predicting the number of new students in the Statistics Study Program at Universitas Negeri Gorontalo and can serve as a reference to improve academic planning quality in higher education institutions.