Science & Technology Asia

Geoinformatics-Based Spatial Evaluation of Durian Cultivation in Chanthaburi, Thailand

2025-08-04T11:21:53+07:00

Sustainable management of high-value perennial crops under climate variability requires reliable spatial tools to assess productivity and constraints. This study used geoinformatics to analyze spatial and temporal variability in durian orchard performance across Chanthaburi province, Thailand, a major production area in Southeast Asia. Multitemporal Sentinel-2 imagery (2022-2025) provided the Normalized Difference Vegetation Index (NDVI), which was integrated with yield per rai and rainfall data from ten districts. NDVI showed a strong positive correlation with yield (𝑟 = 0.74, 𝑝 < 0.01), confirming its reliability as a proxy for crop vigor and productivity. Rainfall had a moderate correlation with NDVI (𝑟 = 0.61) and a weaker one with yield (𝑟 = 0.48). The robust NDVI-yield relationship supports its use in practical yield estimation models when combined with environmental and management factors. These results demonstrate the value of remote sensing for mapping high and low yield zones, enabling precision orchard management, resource optimization, and climate resilience. Overall, the integrated geospatial framework highlights the potential of Earth observation technologies to enhance sustainable orchard practices and advance food security, climate adaptation, and environmental stewardship.

Chemical Content of Sidempuan Salacca (Salacca sumatrana) Vinegar and Its Potential as an Antihypercholesterolemia: an in Vitro Assessment

2025-08-27T11:34:51+07:00

Sidempuan salacca (Salacca sumatrana Becc.) is a snakefruit native to Padangsidempuan, South Tapanuli, North Sumatra, Indonesia, known for its sweet-sour taste. Its sourness reduces market appeal and causes economic losses for farmers, while its perishable nature limits shelf life. To enhance its marketability, processing it into vinegar offers a promising alternative, as salacca vinegar has greater functional potential than apple vinegar. This study aimed to identify the chemical composition of Sidempuan salacca vinegar using GC-MS analysis and evaluate its antihypercholesterolemic potential. GC-MS results revealed five dominant bioactive compounds: oleic acid, DI-(9-octadecenoyl)-glycerol 2-hydroxy-1,3-propanediyl ester, borane, 9-octadecenal, and 9-octadecenoic acid (Z). An in vitro cholesterol-lowering assay using the Liebermann–Burchard method and spectrophotometry showed that the vinegar effectively reduced cholesterol, with an LC50 value of 142.232 ppm. In conclusion, Sidempuan salacca vinegar contains active compounds with antioxidant and anti-cholesterol properties, demonstrating potential as a natural functional product for managing cholesterol levels.

Accuracy of Respiratory Rate Monitoring Using Thammasat Capnoline During Total Intravenous Anesthesia: a Modified Oxygen Nasal Cannula-Based Capnometry Device

2025-05-13T15:37:43+07:00

Continuous respiratory monitoring is crucial for patients receiving total intravenous anesthesia (TIVA), as respiratory depression such as hypoventilation and apnea can be difficult to detect. Capnometry via nasal cannula offers a cost-effective method for continuous
end-tidal carbon dioxide (ETCO₂) monitoring. This study evaluated the correlation between respiratory rates measured by the Thammasat Capnoline device and clinical assessment during TIVA with supplemental oxygen. A prospective observational study was conducted on 63 patients (aged 18–65 years, ASA I–III) undergoing procedures under TIVA. Respiratory rates were simultaneously measured using thoracoabdominal observation and capnometry at various oxygen flow rates. The participants (38.10% male) had a mean age of 40.62 years and mean BMI of 23.46 kg/m², with ASA I (55.56%), II (42.86%), and III (1.59%). Mean respiratory rates (breaths/min) showed no significant difference between methods across oxygen flow rates: 2 L/min (17.12 vs 17.39, 𝑝 = 0.119), 3 L/min (17.35 vs 17.49, 𝑝 = 0.455), 4 L/min (17.32 vs 17.31, 𝑝 = 0.929), and 5 L/min (17.69 vs 17.69, 𝑝 = 0.998). Oxygen flow variation did not affect accuracy (𝑝 = 0.901). Thus, Thammasat Capnoline provides reliable respiratory monitoring comparable to clinical assessment during TIVA.

Solving Disassembly Line Balancing Problems with Fuzzy Parameters Using an Artificial Intelligence Technique

2025-04-17T14:52:42+07:00

End-of-life (EOL) product recovery has become a critical issue driven by economic, social, and environmental concerns, along with stricter environmental regulations that emphasize disassembly and product recovery. Disassembly lines are used to dismantle EOL products into reusable components, but their efficiency depends heavily on accurately estimating task times, which are often uncertain. Since average task times cannot fully represent this uncertainty, task time can instead be modeled as a fuzzy number, allowing fuzzy logic to quantify representative values. This study introduces a disassembly line balancing problem where processing times are expressed as fuzzy numbers and solved using particle swarm optimization (PSO). The optimization aims to minimize the number of workstations, total idle time, maximum disassembly cost, and direction changes. The proposed method was benchmarked against LINGO and GUROBI solvers. Computational experiments, using the number of non-inferior solutions as a stopping criterion, demonstrated that the PSO-based approach achieved superior results. The findings indicate that the proposed method effectively outperforms existing algorithms, providing efficient and promising solutions for EOL product recovery and disassembly optimization.

Predictive Modeling of Brackish Surface Water Quality for Reverse Osmosis Desalination Plants Using Advanced Machine Learning Techniques

2025-07-23T10:28:51+07:00

Artificial intelligence (AI) has proven highly effective in optimizing water treatment processes, particularly for monitoring reverse osmosis (RO) desalination plants treating brackish surface water. These systems face complex, non-linear variations in feed water quality, making real-time monitoring crucial to avoid performance loss and membrane fouling. This study presents a machine learning (ML)–based framework to predict the water quality index (WQI) and enable rapid decision-making by classifying water quality (WQC) into four actionable categories: Excellent, Good, Poor, and Unsuitable. Using 11 key water quality parameters, the model provides an efficient and reliable approach for prediction and classification. Among tested algorithms, the multi-layer perceptron (MLP) achieved the best WQI prediction performance, with an R² of 98.19%, a mean absolute error (MAE) of 0.0182, and a mean squared error (MSE) of 0.0043. For WQC, the XGBoost algorithm outperformed others, reaching 99.84% accuracy. The results demonstrate the strong potential of ML techniques to enhance water quality monitoring and management in RO desalination plants, supporting efficient operation and timely intervention.

Optimization of Submerged Fermentation of Cordyceps militaris Mycelium with Coconut Broth

2025-07-21T14:01:03+07:00

Cordyceps militaris is a well-known medicinal mushroom celebrated for its immunomodulatory, antioxidant, and anti-inflammatory properties. This study explores the optimization of submerged fermentation for C. militaris mycelium using coconut broth as a sustainable nutrient medium. Coconut broth, consisting of coconut sugar, white sugar, and skim milk, was investigated for its effectiveness in promoting biomass yield and bioactive compound production. Using a 2-Level Full Factorial Design and Box-Behnken Design, the study examined varying concentrations of coconut sugar at 9 ◦Brix (100-200 ml/L), white sugar (10-30 g/L), and skim milk (5-10 g/L) to identify the most influential factors affecting the dry weight of C. militaris mycelium (DWCM). Statistical analysis revealed that skim milk, white sugar, and coconut sugar significantly contributed to increased DWCM, with contribution rates of 97.09%, 88.38%, and 74.67%, respectively. The steepest ascent method determined the direction of optimization, and the final regression model identified the ideal conditions for maximum DWCM (27.172 g/L) as 128.38 ml/L of coconut sugar at 9 ◦Brix, 25.476 g/L of white sugar, and 11.466 g/L of skim milk. These results indicate that coconut broth is an eco-friendly and suitable for scaling up production of bioactive compounds and developing innovative functional health products.

Computational Fluid Dynamics Simulation Of T-Pipe Scaling Up With Heavy Oil-Water Core-Annular Flow

2025-08-05T11:27:54+07:00

Core annular flow (CAF) offers significant energy savings for transporting heavy oils, but its large-scale application in complex geometries remains poorly understood. This study investigates the scale-up of CAF in a T-shaped pipe (T50-50) with water insertion using computational fluid dynamics (CFD). The objective is to evaluate the effects of Reynolds number (Re), Bingham number (Bi), and Froude number (Fr) on flow stability, fouling, and energy efficiency. Simulations employed the Volume of Fluid (VOF) method with the Carreau model to capture shear-thinning behavior. Pipe dimensions were scaled by factors of 1.5 and 5, and performance was assessed through pressure gradient, oil holdup, and energy savings. Results show that maintaining Fr > 1 is essential for concentric CAF stability and fouling reduction. At 1.5x scale-up, oil holdup errors relative to theory were 0–19%, while pressure gradient errors were 19–52%. At 5, errors increased sharply, particularly under Reynolds scaling (up to 554%). Despite this, CAF consistently reduced pumping energy by more than 80%, reaching 98–100% savings. In conclusion, CAF remains an effective strategy for heavy oil transport, with Fr governing stability and Bi providing the best predictor of energy efficiency.

Impact of Soil-Structure Interaction on Fragility Curves and Ductility Demands in 3D Reinforced Concrete Moment-Resisting Frames

2025-08-14T10:43:05+07:00

Earthquakes are one of the most devastative natural hazards that cannot be avoided by mankind. Hence, seismic risk mitigation procedures based on vulnerability assessment of the existing infrastructure is the only alternative to prevent the imperative socio- economic and human loss. The seismic analysis and vulnerability assessment procedures of building structures are usually carried out by ignoring the contribution of soil and foundation characteristics in the seismic response of the superstructure. The present work is primarily focused on investigating the soil structure interaction effects on the performance assessment of 4- story and 12-story building configurations with both fixed base and flexible base conditions designed as per IS 456 and IS 1893. Further, fragility analysis is performed using the Capacity Spectrum method and damage probability matrices are then developed for each damage state to describe the structural behavior. It has been observed from the results that consideration of the SSI effect has significantly altered the response characteristics of the models considered. It can also be observed that the constant ‘R’ value specified by IS 1893 for a particular RC MRF was found to be significantly less than the ‘R’ values computed for all the building configurations using NLS analysis. This necessitates the importance of dynamic characteristics of structures in estimating the load-carrying capacity, resulting in an adequate estimation of design forces leading to optimal design configurations.

A Model for Smart Detection: Modified Explainable Machine Learning for Interpreting Detection Results

2025-08-27T11:47:20+07:00

Medical imaging analysis using artificial intelligence has become a powerful system for assisting the doctor in diagnosing some diseases. Most of CAD performed excellent performance with average accuracy of more than 80%. Regardless of the excellent performance of artificial intelligence in the CAD, implementation of artificial intelligence in medical cases is still causing controversy. It happened due to the black-box principle of AI. Actually, both machine learning and deep learning worked in the black-box direction in which it was difficult to recognize how the model performed and how it analyzed the data. Hence, it became controversial since there remained some big questions about “how can the doctor trust the AI result?” Regarding this problem, a continued solution was needed. In this study we proposed a modified SHAP for explaining the artificial intelligence result. The modification itself is conducted by performing correlation in the perturbation process of SHAP. Our proposed solution was performed into two different datasets to evaluate the significance and the reliability of the proposed solution. According to both the visual analysis and statistical test, we conclude that the proposed solution gave a more rational explanation compared to the original SHAP.

Design of an Integrated Modern Explainable Machine Learning Framework for Blockchain Forensic Analysis

2025-08-15T13:51:42+07:00

The rapid expansion of blockchain technology has led to a surge in fraud cases, demanding advanced forensic methods to ensure security and transparency. Traditional static, rule-based models are inadequate for the complex and dynamic nature of blockchain transactions, while existing graph-based anomaly detection methods still struggle with temporal awareness, adaptability, and cross layer integration. These models also suffer from high false positives, inefficient thresholding, and limited explainability, reducing their effectiveness in real-world investigations. To address these gaps, we propose a Modern Explainable Machine Learning Framework for Blockchain Forensic Analysis, integrating five advanced AI techniques to enhance both fraud detection and interpretability. The Temporal Graph Transformer (TGT) identifies evolving transaction patterns with 96.5% accuracy, while Reinforcement Learning-Based Adaptive Fraud Thresholding (RL-FT) reduces false positives by 45%. Contrastive Self-Supervised Blockchain Embedding (CSBE) improves fraud separation by 30%, and Hybrid Diffusion-Based Anomaly Forecasting (HDAF) predicts future fraud with 94.1% accuracy. Finally, Multi-Modal Blockchain Forensic Fusion (MBFF) combines transactional, smart contract, and network data for 99.1% detection accuracy and 50% better explainability. This integrative forensic intelligence system effectively overcomes key limitations in current blockchain fraud analysis.

Post-COVID-19 Online Shopping Behavior Analysis of Thai Consumers: Using FP-Growth

2025-10-10T14:07:09+07:00

This research analyzes customer purchasing patterns on a Thai e-commerce platform using association rule mining. The study employed the FP-Growth algorithm on a dataset of 39,149 transactions from January to August 2023, with a minimum confidence level of 0.7. The analysis yielded 16 significant association rules, revealing complex cross-category purchasing behaviors. Key findings include a 68.6% likelihood of supplement purchases when sports and gift items are bought together, and a 72.9% chance of cosmetics purchases when supplements, appliances, and books are combined. These insights offer valuable guidance for targeted marketing strategies and sales promotion planning in Thailand’s online retail sector. The study contributes to the understanding of e-commerce behavior in the Thai market, with implications for personalized recommendations, user interface optimization, and inventory management.

Disassembly Line Balancing Design for Mixed-Model Disassembly Process Using a Modified Metaheuristic Approach

2025-06-11T10:14:44+07:00

This study has two main objectives: to propose a mathematical model for the mixedmodel disassembly line balancing problem and to develop a customized solving technique. The model aims to minimize the disassembly line length and the number of opened stations while maximizing workload smoothness in a two-sided disassembly line. The solver, based on the particle swarm optimization (PSO) algorithm, was enhanced through a new discretization method and the survival sub-swarm PSO strategy, enabling it to handle multi objective optimization via Pareto optimality for constructing the elite list. To validate the approach, experiments were conducted on a top-loaded washing machine with different takt times (71, 80, 90, and 100 seconds). Four competitive algorithms—NSGA-II, SPEA2, BARON, and MINLP—were used for comparison. Performance was evaluated using three indicators: inverted generational distance (𝐼𝐺𝐷), hypervolume (𝐻𝑉), and ratio (𝑅). The results showed that the proposed method consistently outperformed the other algorithms, achieving superior accuracy, efficiency, and stability in delivering optimal and reliable solutions.

Innovative Techniques for 3D Model Optimization in the Metaverse

2025-08-04T10:32:32+07:00

3D models are essential components in creating a realistic metaverse and providing a good user experience. However, the large file sizes of 3D models often lead to challenges in storage, transmission, and real-time rendering. Therefore, reducing the size of 3D model files during the production process is a key solution to these problems. This research aims to (1) explore methods to optimize 3D models and propose add-ons for 3D modeling software, and (2) compare the effectiveness of these optimization methods. The researchers selected three optimization techniques to evaluate their performance based on key indicators including file size (MB), total vertices, unique materials, batches/draw calls, materials count, and memory usage (KB). The analysis revealed that the Create UV Map Technique combined with Texture Atlas by Merging Material Types into a Single Unit was the most effective method. This technique was then implemented to develop a Blender add-on. The add-on’s performance was tested and compared against models created without it using comparative factors such as file size (KB), mesh vertices, unique materials, batches/draw calls, materials count, and materials memory usage (KB). The comparison showed that models using the add-on had significantly better performance with statistical significance at the 0.01 level.

Comparative Evaluation of Hybrid Deep Learning Models for BISINDO Alphabetic Video

2025-06-10T14:36:08+07:00

Accurately recognizing BISINDO (Bahasa Isyarat Indonesia) alphabetic gestures from video data presents significant challenges due to variations in pose, lighting conditions, and background noise. While previous studies have explored individual deep learning models, such as CNNs or LSTMs, the comparative evaluation of hybrid architectures that combine spatial and temporal feature extraction remains limited, especially in video-based sign language recognition. This study proposes a hybrid deep learning approach that integrates ResNet50 for spatial feature extraction with CNN and LSTM architectures for temporal sequence modeling, aiming to enhance the robustness and accuracy of BISINDO gesture classification. Compared to conventional CNN-based models, our hybrid architecture demonstrated an improvement of 3.4% in F1-score and over 17.6% in precision on challenging gestures. This systematic comparative evaluation reveals the superior capability of hybrid models to generalize in complex environments. Sample videos used in this study contain various backgrounds and signer styles to reflect real-world conditions. The findings contribute to developing more reliable sign language recognition systems and provide insights for future research and practical applications in real-time gesture recognition.

Synthesis and Characterization of Heterogeneous Nano-Catalyst for the Production of Biodiesel from Pongamia Pinnata Oil

2025-05-06T15:09:38+07:00

An alternate energy source for a variety of applications is provided by the production of biodiesel from Pongamia pinnata oil. To load MgO onto the ferrite-coated zirconium oxide, the impregnation method was used. A 0.5 mol solution of magnesium nitrate was allowed to dissolve in distilled water. After adding sodium hydroxide and CTAB, the mixture was added to a solution comprising ferrite- coated zirconium oxide and swirled constantly for eight hours at 65◦C using a hot plate magnetic stirring device. Once the color of the final solution changed, the introduction of sodium hydroxide was halted. The dense, viscous filtrate was cleaned using methanol and distilled water. After that, the thick sediment was dried by being kept at 110◦C for 12 hours in an oven. Lastly, the precursor material was calcined for three hours at 600◦C in a muffle furnace. To investigate its catalytic activity, the produced catalyst was evaluated using FTIR, SEM, XRD, and particle size analysis. A maximum yield of 97 wt% biodiesel was attained, when it was transesterified utilizing the ZrO₂/MgO-Fe₃O₄ catalyst under different conditions: 6% (w/w) catalyst concentration, 15:1 methanol to oil molar ratio, 60◦C reaction temperature, and 240 min reaction duration. The ASTM standard was utilized for the analysis of the fuel qualities. It appears that the synthesized nanocatalyst is quite responsive in terms of both quality and effectiveness.

Regularity of the Transformation Semigroups Restricted by an Equivalence Relation with a Restricted Range

2025-05-20T15:06:11+07:00

Let 𝑋 be a nonempty set. Denote by 𝑃(𝑋) and 𝑇(𝑋) the partial transformation semigroup and the full transformation semigroup on 𝑋, respectively. For a nonempty subset 𝑌 of 𝑋 and an equivalence relation 𝜌 on 𝑋, let
𝑃(𝑋,𝑌, 𝜌) = {𝛼 ∈ 𝑃(𝑋) | ran𝛼 ⊆ 𝑌 and ∀𝑥,𝑦 ∈ dom𝛼,(𝑥,𝑦) ∈ 𝜌 ⇒ 𝑥𝛼 = 𝑦𝛼},
𝑇(𝑋,𝑌, 𝜌) = {𝛼 ∈ 𝑇(𝑋) | ran𝛼 ⊆ 𝑌 and ∀𝑥,𝑦 ∈ 𝑋,(𝑥,𝑦) ∈ 𝜌 ⇒ 𝑥𝛼 = 𝑦𝛼}.
Then 𝑃(𝑋,𝑌𝜌) and 𝑇(𝑋,𝑌, 𝜌) are subsemigroups of 𝑃(𝑋) and 𝑇(𝑋), respectively. In this paper, we characterize the regular elements, left regular elements and right regular elements of both 𝑃(𝑋,𝑌, 𝜌) and 𝑇(𝑋,𝑌, 𝜌). In addition, the regularity, left regularity and right regularity of 𝑃(𝑋,𝑌, 𝜌) and 𝑇(𝑋,𝑌, 𝜌) are determined.

Interval Tolerance Solution for Adjusted Problem of Optimistic Interval Linear Program

2025-06-19T18:14:50+07:00

An optimistic solution to an interval linear program is a real-valued solution derived from the best-case deterministic linear program. However, relying solely on this best-case solution can be overly simplistic, as the actual realization of parameters lies within specified intervals. Instead, it is more appropriate to provide an interval vector solution that remains near the optimistic solution, particularly when the decision-maker prefers proximity to the best-case scenario. In this paper, we establish the equivalence between the weak feasible solution set of an interval equality system and the union of basic feasible solutions across all scenarios of an interval linear program with an interval inequality system, where the interval inequalities require the left-hand side to be lower than the right-hand side, but not excessively so. Furthermore, we demonstrate that, under positive variables, this set coincides with the union of basic optimal solution sets. This result enables the use of a tolerance-based approach to identify an interval vector solution near the optimistic solution. Specifically, we modify the interval linear program so that the optimistic solution becomes a tolerance solution for the adjusted problem. We then propose a method to derive the interval tolerance vector solution for the modified problem, with the goal of maximizing the total sum of the dimensions of the interval tolerance vector hyper-box. Our proposed method differs from most existing methods for finding interval solutions, as those methods typically yield interval solutions that merely include weak solutions without specifying the solution type. Even though there are existing methods for obtaining interval tolerance solutions, none of them consider interval tolerance solutions that are close to the optimistic solution.

HORL_2OPT: A Hybrid Reinforcement Learning and Hippopotamus Optimization Algorithm for Bottled Water Delivery Route Optimization

2025-04-25T13:39:54+07:00

This study presents HORL_2OPT, a hybrid optimization framework developed to address the bottled water delivery routing problem modeled as a Traveling Salesman Problem (TSP). The framework aims to minimize travel distance, enhance computational efficiency, and ensure consistent solutions. HORL_2OPT combines three key components: 𝑄-learning for guided initialization, the Hippopotamus Optimization Algorithm (HOA) for global exploration, and a 2-opt heuristic for local route refinement. Tested on 15 TSPLIB benchmarks and 26 real-world cases from a bottled water distributor in southern Thailand, HORL_2OPT consistently produced the best or near-best results. For instance, it achieved a total distance of 8,034.2 in the berlin52 problem, outperforming HOA (12,953.2), DE (25,215.2), and PSO (23,187.0); and in lin318, it achieved 56,695.0 compared to HOA’s 85,286.2 and DFA’s 122,910.4. In real applications, it generated the shortest or equally optimal routes in 18 of 26 cases, occasionally surpassing LINGO, with most runs completed within 20 seconds. By integrating machine learning, metaheuristics, and local search, HORL_2OPT delivers robust, high-quality solutions suitable for practical logistics and dynamic routing scenarios.

Perimetric Contraction on Quadrilaterals

2025-05-02T14:09:14+07:00

In this article, we introduce a four-point analogue of the Banach contraction principle and establish sufficient conditions for such mappings to possess fixed point(s) in complete metric spaces. Notably, the classical Banach contraction principle emerges as a special case of our results. We present several non-trivial examples that not only validate our theorems but also reveal that quadrilateral perimetric contractions need not imply other well-known contraction types. Furthermore, we extend our analysis to obtain fixed point theorems in non-complete metric spaces. Lastly, we address a recent result linking mappings contracting the perimeters of triangles in metric spaces to Banach-type contractions in 𝐺-metric spaces.

Assessment of the Potential of Sodium Benzoate as Anti-Microbial and Anti-Corrosion Agent in Coolant

2025-07-02T09:22:39+07:00

Sodium benzoate, as both an antimicrobial agent and corrosion inhibitor, is ideally suited for application in engine coolant. This study evaluated the antimicrobial and anticorrosion properties of sodium benzoate in an ethylene glycol-based organic engine coolant. Antimicrobial assays tested sodium benzoate against Bacillus cereus, Pseudomonas aeruginosa, and Trichoderma viviens. The corrosion inhibitory effect was assessed using the ASTM D1384 glassware corrosion test. It was found that 3% wt sodium benzoate strongly inhibited microbial growth, while a minimum dosage of 1% wt provided only mild antimicrobial performance. Interestingly, the presence of 50% ethylene glycol in the coolant outperformed sodium benzoate in antimicrobial effectiveness. Sodium benzoate also exhibited selective corrosion protection; at a dosage of 1% wt, it offered superior protection for aluminum but promoted corrosion of iron.

Prediction of Indonesian Coal Prices Using Support Vector Regression Method with Grid Search Optimization

2025-05-18T11:31:45+07:00

Coal prices are highly volatile due to fluctuations in global supply and demand, government policies, and economic and political factors. Therefore, accurate price prediction is essential and can be achieved using Support Vector Regression (SVR). This study aimed to identify the most accurate SVR model for forecasting Indonesian coal prices and analyze its prediction performance. The Radial Basis Function (RBF) kernel was used with hyperparameter ranges of 10⁻³ ≤ 𝐶 ≤ 10², 10⁻³ ≤ 𝛾 ≤ 10², and 𝜀 = 0.01, 0.02, 0.03, tested at data splits of 70:30, 80:20, and 90:10. Optimal hyperparameters were determined using the grid search algorithm, and model performance was evaluated using the Mean Absolute Percentage Error (MAPE). Results showed that the best prediction accuracy was achieved with a MAPE of 5.459%, indicating excellent model performance. The optimal configuration was the SVR with RBF kernel at a 90:10 ratio, where 𝐶 = 12.5, 𝛾 = 0.1, and 𝜀 = 0.025. The resulting model effectively approximated the actual coal price data, confirming its reliability for forecasting Indonesian coal prices.

On the Convergence and Stability of a Hybrid Iteration Scheme in Uniformly Convex Banach Space

2025-06-19T18:47:08+07:00

A new iterative algorithm for approximating fixed points is considered on the lines of the iterative algorithm considered by Pansuwan and Sintunavarat [1]. The convergence of the considered iterative algorithm is established. Finally, the convergence rate of the new
iterative algorithm is compared with that of the iterative algorithm considered in [1].

Strong Homomorphisms of Topological Groupoids

2025-08-15T13:23:48+07:00

In this work, we generalized to topological groupoids that we known the strong morphism for groups. Also we obtained some characterizations of these homomorphisms. Namely, when there is a homomorphism between algebraic structures, there is an important result that arises from this homomorphism between the substructures of these algebraic structures: the correspondence theorem between substructures. From the perspective of groupoid theory, this theorem establishes a bijective correspondence between the subgroupoids of two related groupoids under a surjective homomorphism.

Interval Estimation in Truncated Spline Regression: Analyzing the 2023 Indonesian Democracy Index

2025-07-14T14:18:53+07:00

The truncated spline is a piecewise polynomial that maintains continuity across segments and offers a high degree of flexibility in estimating data that varies significantly across different intervals. This study applies the truncated spline method with interval estimation to analyze the 2023 Indonesian Democracy Index (IDI) data. The objective is to identify the key factors that significantly influence IDI and to evaluate the estimated results of the nonparametric spline regression curve for the 2023 IDI data. The analysis findings indicate that the predictor variables used in this study have a substantial impact on IDI, with a determined coefficient of approximately 97.81%. Based on the interval estimation analysis of the regression curve, it was found that out of 34 provinces in Indonesia, 8 provinces were estimated accurately. These provinces include Kepulauan Riau, DKI Jakarta, Bali, Central Kalimantan, North Sulawesi, Central Sulawesi, North Maluku, and Papua. Additionally, the analysis indicates that 16 provinces have experienced a decline in their Democracy Index (IDI) criteria, suggesting a possible transition in their democratic performance. This situation calls for further in-depth studies to improve their democratic achievement levels. On the other hand, 18 provinces are projected to experience an increase in their IDI performance.

Analyzing PM2.5 Levels in Indonesia Using Dynamic Time Warping and Fuzzy Clustering: A Time-Series Ecological Study

2025-10-15T10:07:59+07:00

Air pollution from particulate matter (PM2.5) poses serious global health risks. This study evaluated five fuzzy clustering methods Fuzzy C-Means (FCM), Fuzzy Possibilistic C-Means (FPCM), Possibilistic Fuzzy C-Means (PFCM), Fuzzy Gustafson-Kessel (FGK), and Fuzzy C-Shells (FCS)—integrated with Dynamic Time Warping (DTW) to cluster PM2.5 levels across 33 Indonesian capital cities. Using an ecological time series design, daily PM2.5 data (March 6, 2023–March 5, 2024; 12,045 data points) from PlumeLabs were analyzed in R 4.4.2 with descriptive and inferential statistics, including the Kolmogorov-Smirnov, Kruskal-Wallis, and Dunn tests. Three clusters—high, moderate, and low pollution—were identified. The DTW+FCS method showed the best performance (PCI: 0.798, MPCI: 0.697, PEI: 0.357, XBI: 0.197) with significant differences (𝑝 < 0.000). These findings highlight DTW+FCS as the optimal approach and emphasize targeted air quality strategies for Indonesia’s highpollution areas.

Research on Optimizing Control Signals for Single-Agent Navigation under Multiple Scenarios in Linear Motion

2025-03-04T13:50:49+07:00

This study proposes a Bezier curve optimization method for enhancing the control signals and trajectory tracking performance of the Donkey Car, a 1:16 scale autonomous vehicle platform. The method employs second-order Bezier curves for throttle optimization and third-order Bezier curves for steering angle optimization. A novel loss function, Bezier Smoothing Loss (BSL), is introduced to simultaneously optimize control signal smoothness and trajectory tracking accuracy during neural network controller training.Experiments in three scenarios (left lane driving with obstacle avoidance, straight line driving, and straight driving with continuous obstacle avoidance) show that the proposed method significantly improves trajectory tracking accuracy (RMSE reduced by up to 19.6%), control signal smoothness (throttle change rate standard deviation decreased by 28.6%, steering angular velocity standard deviation reduced by 24.5%), and vehicle posture stability (yaw rate and pitch rate RMS values decreased by 7.9%). Compared to other learning-based methods (KerasLinear, KerasRNN, PBLM-CNN21, MFPE-CNN14), our approach achieves superior performance across all evaluation metrics.The proposed Bezier curve optimization approach effectively refines the performance of autonomous driving systems and offers a promising direction for future research and development in this domain.

The Inverse Moyal Distribution and Its Properties

2025-07-07T14:49:02+07:00

In this paper, we introduce a new extension of the Moyal distribution, called the inverse Moyal (IMoyal) distribution, for heavy-tailed distribution. We derive its important statistical properties, such as the probability density function, cumulative distribution function, survival function, hazard function, and cumulative hazard function. Furthermore, we provide the formula for maximum likelihood estimation and perform simulation studies to assess its performance. Finally, we apply the IMoyal distribution to two real-world datasets: an automobile insurance premium data and a commercial bank asset data of Thailand. The analyses show that the proposed distribution demonstrates a superior fit to the data compared to existing distributions.