This study empirically investigates actual customer repurchase behavior by analyzing behavioral log and payment data collected from study cafe users in an attendance-based learning service environment. Unlike prior studies that primarily relied on survey-based measures, this study integrates attendance records, payment data, and explainable artificial intelligence (XAI) techniques to provide a data-driven understanding of customer retention behavior. It compares the predictive performance of traditional regression models with machine learning approaches. Specifically, logistic regression, random forest, XGBoost, and neural networks were employed, with SHAP analysis applied as an XAI technique. The results indicate that visit frequency has a significant positive effect, supporting H1, while stay-duration variables show only limited and inconsistent effects, providing partial support for H2 and H3. Total payment in May negatively affects subsequent repurchase, suggesting possible saturation or substitution effects, thereby supporting H4. Age demonstrates a negative effect (supporting H5), and regional differences are captured by the XGBoost model (supporting H6). In terms of predictive performance, machine learning models outperformed logistic regression, with XGBoost achieving the strongest overall results among the evaluated models (ROC-AUC = 0.676; PR-AUC = 0.642). Overall, this study contributes to the literature by presenting empirical evidence based on behavioral data and by highlighting the practical interpretability of integrating XAI techniques. From a managerial perspective, the findings provide actionable insights for designing customer retention strategies based on visit frequency, spending behavior, and regional characteristics, thereby supporting AI-driven decision-making in attendance-based service environments.
Indonesia’s palm oil downstreaming strategy has expanded domestic processing, product diversification and bioenergy utilization, but its distributional architecture remains incomplete because smallholders continue to participate primarily as suppliers of fresh fruit bunches (FFB). This article develops a smallholder-centred interpretation of downstreaming through a qualitative integrative literature review using narrative and thematic synthesis. The review draws predominantly on peer-reviewed literature published from 2020 to August 2026 and uses recent Indonesian policy documents only to contextualize current policy directions. The synthesis shows that smallholder downstreaming should not be equated with universal farmer ownership of palm oil mills; instead, it is better understood as progressive value-chain upgrading from productive and legal farming to collective marketing, traceability, logistics, processing participation, downstream small and medium enterprises, circular-bioeconomy activities and, where capabilities permit, equity participation in larger industrial ventures. Five interdependent constraints—productivity and replanting, institutional capability, finance, market power, and sustainability compliance—limit this transition. The article proposes an eight-pillar policy architecture that integrates sustainable intensification, land and farm formalization, cooperative professionalization, differentiated finance, equitable partnerships, scale-appropriate processing, circular bioeconomy development and market creation. Its conceptual contribution is to distinguish national downstreaming success from inclusive downstreaming success by emphasizing who acquires capabilities, who bears risks and compliance costs, and who captures value after FFB leaves the farm. The analysis concludes that Indonesia can strengthen industrial competitiveness and rural development simultaneously only if downstream policies transmit resources, economic rights and value-capture opportunities upstream to smallholders.
The study assessed the current status of hydraulic ram pump utilization in the Philippines. This developed a georeferenced map showing the location of established hydraulic ram pump installations and described the spatial patterns of ram pump installations. A systematic review and meta-analysis approach guided by PRISMA 2020 standards was employed to collect and synthesize data from peer-reviewed publications, institutional reports, and online sources. Web-based data collection and geocoding techniques were used to determine installation locations, which were then analysed through geospatial and artificial intelligence–assisted interpretation. A total of ninety-three (93) hydraulic ram pump installations were identified nationwide. Results revealed that adoption is unevenly distributed and concentrated primarily in regions with favourable topography, watershed availability, and rural water demand. Regions with mountainous and upland characteristics exhibited higher utilization, while highly urbanized areas and regions with institutional and infrastructural constraints showed minimal or no installations.
This study proposes a graph-based framework for anomaly detection and future attack prediction by redefining Digital Identity (DI) as a dynamic and structural security entity rather than a simple authentication credential. The proposed model integrates user, device, network, session, and behavioral information to construct a unique digital fingerprint and represents the relationships among DI components as a graph structure. Normal DI maintains consistent structural patterns, whereas attack situations introduce changes in relational structures and behavioral transition patterns, which are utilized as key indicators for anomaly detection. In addition, temporal correlation analysis is incorporated to extend the framework toward proactive prediction of potential future attacks. Experimental evaluations are conducted using public cybersecurity datasets and DI-based synthetic attack scenarios, demonstrating that the proposed framework achieves higher detection accuracy and improved structural interpretability compared with conventional anomaly detection methods. This study is meaningful in that it reinterprets DI as an active security entity capable of continuous integrity verification, anomaly detection, and future attack prediction.
Indonesia's mandatory B50 biodiesel programme, officially launched in July 2026, represents one of the world's most ambitious attempts to substitute petroleum diesel with a domestically produced biofuel. The policy is intended to strengthen energy security, reduce diesel imports, deepen palm-oil downstreaming, support rural livelihoods, and lower greenhouse-gas emissions. Yet moving from B40 to B50 changes more than the blending ratio. It increases the coupling between the national energy system and the palm-oil economy, thereby intensifying interactions among feedstock availability, plantation productivity, smallholder replanting, crude palm oil (CPO) allocation, commodity prices, biodiesel quality, storage and distribution infrastructure, engine compatibility, fiscal support, environmental safeguards, traceability, and international trade rules. This article conducts a qualitative integrative literature review, rather than a systematic literature review, drawing primarily on peer-reviewed studies published from 2020 to 2026 and triangulating them with current government sources. The review identifies nine interconnected challenge domains and argues that the principal risk is not any single bottleneck but the migration of bottlenecks across the upstream-to-downstream system. B50 can improve energy resilience, but its durability depends on productivity-led feedstock growth, smallholder-inclusive replanting, end-to-end fuel-quality assurance, resilient financing, credible lifecycle sustainability, and adaptive governance. The article proposes a policy architecture in which the mandate is supported by monitored thresholds for feedstock, prices, fiscal exposure, technical performance, distribution reliability, and environmental integrity.
