This study is based on the increase in Gross Regional Domestic Product (GRDP) and the decrease in poverty rates in 2024. This indicates a relationship between the two variables. Previous studies have shown the influence of poverty rates on GRDP and vice versa, these relationships were generally examined using single-equation models that did not account for possible simultaneous interactions. Therefore, this study aims to analyze the relationship between poverty rates and GRDP using a simultaneous equation model with The Two Stage Least Squares (2SLS) approach to see the two-way relationship of the two variables. The results show that GRDP has a significant effect on poverty rates, while poverty rates do not have a significant effect on GRDP. These findings provide empirical evidence that the relationship between the two variables is not reciprocal but rather unidirectional, with GRDP being the only variable that significantly influences poverty rates.

This study models the volatility of daily returns for PT Aneka Tambang Tbk (ANTM), which exhibits volatility clustering, fat tails, asymmetry, and long memory. The analysis proceeds in two stages: (i) conditional variance modeling using FIGARCH (1, d, 1) to represent long-memory dynamics; and (ii) design of a FIGARCH ANN hybrid (backpropagation) to absorb residual nonlinearity/asymmetry. Preprocessing tests confirm stationarity in log returns, followed by ARIMA baseline selection and confirmation of conditional heteroskedasticity in the residuals. Long-memory estimation via the GPH procedure confirms statistically significant long memory (past shocks have persistent effects). Compared to GARCH (1,1) and EGARCH, FIGARCH provides a better fit because it has the smallest AIC/BIC value. Residual diagnostics for FIGARCH are clean, indicating no autocorrelation and no remaining ARCH effects the model captures the main volatility structure. Ten steps ahead forecasts show the conditional variance stabilizing, implying that shocks decay slowly (persistence) toward a relatively stable level. The ANN component trained on residuals/logvariance reduces error metrics (MSE/RMSE/MAE) compared with standalone FIGARCH, evidencing the benefit of nonlinear correction for short horizon accuracy.

The escalating complexity of cyber-attacks and the severe consequences of data breaches necessitate a shift toward more advanced, yet accountable, network security infrastructures. While deep learning models offer superior performance in identifying intrusions, their inherent ”black-box” nature hinders practical adoption in critical environments where security analysts must verify alerts, justify defensive actions, and conduct forensic audits. To address this lack of transparency, this study proposes an explainable Deep Neural Network (DNN) framework for a reliable Intrusion Detection System (IDS) using the CIC-IDS2017 dataset. By integrating the Shapley Additive Explanations (SHAP) method, we bridge the gap between high-performance detection and interpretability. Our experimental results demonstrate that our minimalist DNN architecture achieves an outstanding accuracy of 99.57% and an AUC-ROC of 0.9997, maintaining high detection rates across various attack types with significantly lower computational overhead compared to complex hybrid models. Furthermore, the SHAP analysis identifies Flow IAT Std and Packet Length Variance as the most influential features, offering granular insights into the model’s reasoning. This research demonstrates that high-performance deep learning can be paired with rigorous interpretability, providing a robust and transparent solution for real-time network security monitoring.

Early fault detection in rolling element bearings remains a challenging problem, particularly under unsupervised conditions where labeled fault data are unavailable. Incipient defects often generate weak impulsive vibration signatures that are easily masked by operational noise. This paper proposes a residual-based unsupervised fault detection method that integrates Independent Component Analysis (ICA) with a Long Short-Term Memory (LSTM) autoencoder. ICA is employed to decompose vibration signals into statistically independent components, enhancing fault-related impulsive features while suppressing redundant background vibration. An LSTM autoencoder is trained exclusively on healthy-condition data to learn normal temporal dynamics. Bearing anomalies are identified through reconstruction residuals, which quantify deviations from learned healthy behavior. Instead of fixed or heuristic thresholds, the decision boundary is determined via F1-score optimization, framing fault detection as a data-driven residual decision problem. The proposed approach is validated using the Case Western Reserve University (CWRU) bearing dataset under a 2 hp load condition, focusing on inner race faults. Experimental results demonstrate perfect fault recall and an ROC-AUC of 1.000, confirming the effectiveness of ICA-enhanced residual learning for early fault detection. The method is computationally efficient, interpretable, and suitable for practical predictive maintenance applications.

study proposes a hybrid anomaly detection framework based on Independent Component Analysis (ICA) and Bayesian Long Short-Term Memory (LSTM) networks for early fault diagnosis in rolling element bearings. The model is trained exclusively on normal condition data from the Case Western Reserve University (CWRU) Bearing Dataset, enabling unsupervised deployment suitable for real-world industrial environments. Vibration signals are preprocessed using ICA to extract statistically independent features, followed by a Bayesian LSTM that captures temporal dynamics and provides uncertainty-quantified predictions. A dynamic thresholding mechanism based on Interquartile Range (IQR) autonomously distinguishes between normal and anomalous behavior without manual calibration. Experimental results demonstrate exceptional performance with 99.1% accuracy, perfect recall (100%) on fault conditions, and zero false negatives, ensuring no faults are missed. The composite anomaly score effectively tracks degradation progression, thereby rendering the system highly reliable and practical for predictive maintenance applications. This approach combines statistical signal processing with probabilistic deep learning to deliver a robust, explainable, and adaptive solution for bearing health monitoring.

Type of the article: Research Article AbstractRegional transfers are a key instrument of regional fiscal policy that promote balanced development and reduce disparities across districts, yet the behavioral responses of local governments to these inflows remain insufficiently understood. This study examines how regional transfer fund allocations influence government expenditure, unemployment, infrastructure development, and regional revenue in West Sumatra using quarterly data for 2014–2024. A Bayesian Vector Autoregressive framework is employed to address small-sample limitations and to capture the dynamic responses to transfer shocks. The results show that increases in regional transfers have limited, short-lived effects on unemployment, while capital expenditure on basic infrastructure declines, indicating potential crowding-out of certain government spending categories. At the same time, regional revenue responds positively, suggesting that transfers can support local fiscal capacity in the short term. These findings highlight that, although regional transfers can facilitate immediate fiscal stabilization, they may hinder long-term infrastructure investment unless accompanied by performance-based fiscal mechanisms. Improving transfer design and accountability is therefore essential to ensure that fiscal resources promote sustainable development outcomes. AcknowledgmentsThis research is a grant from the Ministry of Higher Education, Science, and Technology of the Republic of Indonesia under the Impactful Leading Consortium Research Scheme (RIKUB Scheme) in accordance with research contract number 008/C3/DT.05.00/RIKUB/2025.

In Industry 4.0, reliable fault diagnosis is critical for minimizing downtime and preventing catastrophic failures in rotating machinery. However, conventional deep learning models often operate deterministically, lacking the ability to quantify prediction uncertainty—a limitation that hinders risk-based maintenance decisions. This study aims to develop a hybrid deep learning framework that integrates Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and Bayesian inference for uncertainty-aware fault diagnosis. The model extracts spatial features from Short-Time Fourier Transform (STFT) spectrograms via CNN, models temporal dynamics from raw vibration signals via LSTM, and quantifies prediction uncertainty using Monte Carlo Dropout (T=50). Evaluated on the benchmark Case Western Reserve University (CWRU) bearing dataset with an 80/20 data partitioning under six operating conditions, the hybrid architecture achieves an accuracy of 99.14% and an F1-score of 0.9914, significantly outperforming standalone CNN (97.42%) and LSTM (84.12%) models. The integration of probabilistic inference enhances decision reliability by providing confidence estimates for each prediction. This work contributes a robust, uncertainty-aware model that effectively captures both spatial and temporal patterns, offering significant implications for safety-critical industrial predictive maintenance systems.

Almost all cities in Indonesia are moving to find alternatives for providing clean water as the population increases, so the demand for clean water volume is also increasing. Water wastage issues arise as water production continues to rise, highlighting the need for accurate predictions of future water production levels. Modeling clean water supply structural movements as an effort to predict water needs in a city in the future is important. This can be done by considering water usage data over time and factors that trigger clean water demand. This research proposes a long short term memory (LSTM) model that adopts a neuro informatics model as the neural networks approach for modeling water supply. The architecture of the LSTM used in this research employs one hidden layer with 32 neurons. The findings demonstrate that LSTM model can predict water production levels accurately with mean absolute percentage error (MAPE) less than 5% both for training and testing data set. These results categorize the LSTM model as a reliable forecasting tool for water production levels. Therefore, modeling using the LSTM method is a preferable choice for predicting water production aiding relevant parties in planning clean water resources tailored to the needs of population.

This study aims to model and forecast the volatility of ANTM stock returns using FIGARCH and GARCH models to capture both short- and long-memory dynamics. Daily return data spanning from January 1, 2014, to December 31, 2024, were analyzed after stationarity confirmation via ADF test. A mean model was estimated using MA (4), followed by conditional variance modeling with GARCH (1,1) and FIGARCH (1, d,1). Diagnostic tests confirmed the presence of heteroskedasticity and long memory, justifying FIGARCH usage. The FIGARCH (1, d,1) model indicated significant long-memory effects (d = 0.461007), while GARCH (1,1) effectively captured short-term volatility clustering. Forecast performance comparison showed that although both models yielded equal RMSE (0.029000), GARCH (1,1) performed better in terms of MAE (0.019531 vs. 0.019529) and MAPE (192.0809 vs. 192.3617). However, FIGARCH demonstrated superior ability in modeling persistent volatility patterns with smoother conditional variance distribution and better long-term uncertainty estimation. These findings suggest that while GARCH is preferable for short-term predictive accuracy, FIGARCH offers more robust insights into long-term volatility persistence, making it suitable for strategic financial risk management. Penelitian ini bertujuan untuk memodelkan dan meramalkan volatilitas return saham ANTM menggunakan model GARCH dan FIGARCH guna menangkap dinamika volatilitas jangka pendek dan panjang. Data return harian dari 1 Januari 2014 hingga 31 Desember 2024 dianalisis setelah melalui uji stasioneritas ADF. Model rata-rata ditentukan menggunakan MA (4), dilanjutkan dengan pemodelan varian bersyarat menggunakan GARCH (1,1) dan FIGARCH (1, d,1). Uji diagnostik menunjukkan adanya heteroskedastisitas dan efek memori panjang, mendukung penggunaan model FIGARCH. Hasil estimasi menunjukkan bahwa model FIGARCH (1, d,1) memiliki nilai d = 0,461007, mengindikasikan adanya efek long memory yang signifikan, sedangkan GARCH (1,1) efektif dalam menangkap klaster volatilitas jangka pendek. Evaluasi kinerja peramalan menunjukkan kedua model memiliki nilai RMSE yang sama (0,029000), namun GARCH (1,1) lebih unggul dalam MAE (0,019531 vs. 0,019529) dan MAPE (192,0809 vs. 192,3617). Meskipun demikian, FIGARCH menunjukkan keunggulan dalam menangkap pola volatilitas jangka panjang yang stabil. Dengan demikian, GARCH cocok untuk akurasi prediksi jangka pendek, sementara FIGARCH lebih direkomendasikan untuk estimasi risiko jangka panjang dalam pengelolaan keuangan strategis.

Intensification and high external inputs in agriculture can increase global food production. However, this approach can also harm the environment, create social conflict, and make farmers more vulnerable to input prices. A sustainable agriculture system is needed to balance production, the environment, and social systems. Strengthening farmer capacity is crucial in agricultural systems because it helps farmers build sustainable systems. This study aims to formulate sustainable sago plantation development through enhancing farmers capacity. The study was conducted in Indragiri Hilir Regency from November 2023 to January 2024. Data was collected through observation, focus group discussion, and in-depth interviews. Informants in this study consisted of farmers, community leaders, sago harvester, sago traders, factory owners, agricultural extension workers, and the Indragiri Hilir Regency Plantation Office. Data analysis was conducted qualitatively through thematic analysis. The results of the study indicate that sago plantations and their processing industries produce abundant biomass that has the potential to be a source of organic matter, nutrients, and animal feed. Returning organic matter to the land can reduce the cost of using artificial chemical fertilizers, create healthy soil conditions, and reduce environmental pollution. Farmer capacity plays a role in optimizing the use of organic materials as inputs for sago plantations. Farmer capacity building is achieved through innovation, increased local knowledge, and strengthening local institutions. Local institutions provide a platform for farmers to learn together and make their work processes more efficient.

Climate change is one of the major challenges in the world today, characterized by changes in meteorological values, such as rainfall and temperature, caused by the concentration of greenhouse gases in the atmosphere, such as CO2, N2O, and CH4. These accumulated greenhouse gases form a layer that prevents heat radiation from escaping, causing the greenhouse effect and global warming. Addressing the effects of greenhouse gas emissions requires appropriate strategies, one of which is to predict future greenhouse gas emissions for planning appropriate actions. Time series models such as the Autoregressive Integrated Moving Average (ARIMA) model are often used but have drawbacks due to their assumption of linear relationships. On the other hand, the Long Short-Term Memory (LSTM) model, introduced by Hochreiter and Schmidhuber in 1997, can learn complex and nonlinear relationships in data. This study uses LSTM to estimate greenhouse gas emissions in Indonesia based on emitting sectors, hoping to anticipate negative impacts and reduce greenhouse gas emissions. The results show that the LSTM model has good performance with an error below 20%, and it is predicted that greenhouse gas emissions will continue to increase.

In this study, the best parameter estimator for the scale parameter (B) of the inverse Rayleigh distribution was determined based on a comparison of the maximum likelihood estimator (MLE) method, the Bayesian generalized squared error loss function (SELF), the Bayesian linear exponential loss function (LINEX LF), and the Bayesian entropy loss function (ELF). The prior distribution chosen was the non-informative prior, namely the Jeffrey prior, and the informative prior using the exponential distribution. The estimator evaluation method used was based on the smallest value of the Akaike information criterion (AIC), corrected Akaike information criterion (AICc), and Bayesian information criterion (BIC). Based on simulation studies and real data, it was found that the best parameter estimator on the data for the scale parameter (B) of the inverse Rayleigh distribution is the Bayes ELF prior exponential( B EE ).

Stunting is a child who has a height that is shorter than the age standard. One of the main indicators of stunting is a height that is lower than the standard for toddlers. Stunting in Indonesia is of great concern due to the high prevalence of stunting. Stunting children are at risk of impaired cognitive development, which will result in the development of human resources. This study aims to develop a classification model to detect stunted toddlers based on height using the Bayesian binary quantile regression method with LASSO (Least Absolute Shrinkage and Selection Operator). This method was chosen because of its ability to handle multicollinearity and variable selection problems automatically, as well as provide better estimates on non-normally distributed data. The data used in this study includes five independent variables such as age, weight at birth, gender, how to measure height and nutritional status. The results showed that independent variables that significantly affect the height of stunting toddlers can be a concern to reduce the problem of stunting in Indonesia. The results of model show that variable age, weight at birth, and nutritional status have a significant influence to classification of stunting toddler height. Indicator of model goodness is seen from the quantile that has the smallest MSE value. The model that has the smallest MSE is in quantile 0.25 with an MSE value of 0.1622.

The development of tidal swamp land for sago plantations is a potential source of food for future generations. Sago is a native plant of Indonesia that has the highest carbohydrate content and can grow properly in flooded conditions. Through management considering local wisdom, sago plantations in Indonesia have survived for several years despite limited infrastructure, global climate change, lack of government policy support, and changes in community food preferences for rice and wheat. Therefore, this study aimed to determine the effect of management practices on the sustainability of sago plantations and analyze local wisdom values. The research was conducted from November 2023 to January 2024 in Indragiri Hilir Regency, Riau Province, because sago has been planted in this area for hundreds of years with locally wise management. A mixed method with explanatory sequential design was employed, namely a combination of quantitative analysis with the Structural Equation Modeling - Partial Least Squares (SEM-PLS) and qualitative analysis with the Miles-Huberman approach. The results showed that local wisdom was in the form of local knowledge, socio-spiritual values, and community norms. This local wisdom existed from the pre-planting to the marketing of the harvest. Furthermore, local knowledge as well as social values and norms had a significant effect on sustainability, but current harvesting and water management practices are not optimal. The results suggested that practices in water management and harvesting required innovation to be more effective and increase plantation productivity through collaboration with science-based knowledge and formal regulatory systems. Keywords: local knowledge; scientific local wisdom; sago sustainability.

This research investigates the management of CO₂ emissions, a significant factor in the climate change phenomenon, focusing on Indonesia. The objective is to examine the correlation between CO₂ emissions and their causal variables: economic growth (measured by gross domestic product), forest area, and renewable energy (RE) consumption. The Bayesian vector autoregressive (BVAR) model was employed to address the complexity of multivariate interactions and overcome limitations associated with small datasets. The analysis revealed that economic growth and reduced forest area significantly contributed to high CO₂ emissions, while renewable energy consumption exhibited a mitigating effect. The BVAR model demonstrated substantial predictive accuracy, highlighting its suitability for analyzing environmental and economic data in resource-constrained scenarios. These findings emphasize the critical need for targeted policy actions in Indonesia, including safeguarding forest areas, addressing illegal logging and burning, and accelerating the transition to renewable energy. The study provides a novel application of the BVAR model in environmental research, showcasing its potential for generating actionable insights into emissions management. This study contributes to the understanding of sustainable development by proposing an innovative way to support evidence-based policies that reduce CO₂ emissions as well as mitigate climate change impacts.

This study aims to develop a predictive maintenance system for an aging vertical grinding machine, operational since 1978, by integrating machine learning techniques, vibration analysis, and fuzzy logic. The research addresses the challenges of increased wear and unexpected failures in older machinery, which can lead to costly downtime and reduced operational efficiency. Vibration and temperature data were collected over 12 days using an MPU-9250 accelerometer, with conditions categorized as good, fair, and faulty. Various machine learning models, including logistic regression, k-nearest neighbors, support vector machines, decision trees, random forest, and Naive Bayes, were trained to classify bearing states. The random forest model achieved the highest accuracy of 94.59%, demonstrating its effectiveness in predicting machine failures. The results highlight the potential of combining multi-dimensional sensor data with advanced analytics to enable early fault detection, minimize downtime, and improve operational efficiency. This approach provides a cost-effective solution for maintaining aging machinery and contributes to both theoretical advancements in machine learning applications and practical improvements in industrial maintenance practices. The study’s findings offer scalable insights for industries reliant on legacy equipment, promoting sustainable manufacturing through optimized resource use and enhanced reliability.

The poverty line is the threshold income level below which a person or household is considered to be living in poverty. The poverty line is a representation of the minimum rupiah amount needed to meet the minimum basic food needs equivalent to 2100 kilocalories per capita per day and basic non-food needs. According to data from the Central Bureau of Statistics (BPS), although the poverty rate in West Sumatra has decreased in recent years, the issue of poverty is still very relevant to be discussed and addressed. The issue of the poverty line is important to discuss because it is directly related to the welfare of people and the development of a country. For modeling the poverty line and its influencing factors, appropriate statistical methods are needed. This research is about the comparison of two methods, namely the Bayesian quantile regression method and Bayesian LASSO quantile regression. The two methods are compared with the aim of seeing which method produces the smallest error. Bayesian quantile regression is one method that can model data assuming heteroscedasticity violations. This study compares the ordinary Bayesian quantile regression method with penalized LASSO. These two methods are applied in modeling the poverty line in West Sumatra. The purpose of this study is to see the best method for modeling data. The data used amounted to 133 data points from BPS in the years 2017 and 2023. Model parameters were estimated using MCMC with a Gibbs sampling approach. The results show that the Bayesian LASSO method is superior to the method without LASSO. This is evidenced that the superior method produces the smallest MSE value, 0.208, at quantile 0.5. Model poverty line in West Sumatra is significantly influenced by per capita spending ), Gross Regional Domestic Product ), Human Development Index ), Open Unemployment Rate , and minimum wages .