Trisakti University

Accurate prediction of temperature and relative humidity is essential in Solar Dryer Dome (SDD) facilities to ensure strict quality of the agricultural products. Compared to traditional control models, ML-based approaches provide benefits in terms of adaptability, as they are capable of capturing complex and non-linear patterns in time series data. In this project, primary time series datasets collected from SDD facility in North Jakarta were used to forecast the next hour temperature and relative humidity values within the 5-minutes resolution. There were three sensors placed in three different locations: outside the facility near the front door, inside the facility near the front door, and near the fan inlet. The collected data showed daily seasonality patterns with temperature and relative humidity possess an inversely correlated relationship due to their inherent physical interaction. On the modeling part, this research compared three different predictive models such as XGBoost, Linear Regression, and Facebook's Prophet based on the RMSE and R2 scores on the test set. XGBoost and Linear Regression provided on-par performance, with XGBoost showing its best at temperature prediction (RMSE: 1.47 degrees C, R2: 0.96) and Linear Regression showing its best at relative humidity prediction (RMSE: 3.20%, R2: 0.96). Meanwhile, Prophet exhibits the lowest performance in both variables with RMSE: 3.56 degrees C and R2: 0.72 for temperature, and RMSE: 8.42% and R2: 0.76 for relative humidity forecasting. Despite the low performance, the Prophet model was advantageous in terms of consistency across horizons, while the two best models show decreasing performance in higher horizon settings. These findings suggest the promising application of ML techniques in SDD facilities to support sustainability in agroindustrial business.

The conventional health insurance sector faces challenges in accurately evaluating risk and pricing policies, as it primarily relies on aggregated demographic data and general medical histories. This often leads to inefficient premium structures and limited preventive care opportunities, highlighting the need for more individualized risk assessment. This study aims to examine how Big Data Analytics (BDA) and machine learning can enhance the design of personalized health insurance products by integrating real-time, individual-level data beyond traditional actuarial methods. The method involves analyzing a large dataset of 50,000 policyholders over three years, including anonymized electronic health records (EHRs), wearable device data (activity and vital signs), social determinants of health (SDOH), and claims history. Predictive modeling was conducted using XGBoost and Random Forest to estimate individual-level claim frequency and severity. The results show that the BDA-driven approach outperforms traditional actuarial models, achieving an AUC of 0.89. Key predictors of high-cost claims include sleep quality and heart rate variability. These insights enable the creation of hyper-segmented insurance products with dynamic premiums and behavior-based incentives. In conclusion, integrating BDA into health insurance underwriting improves pricing accuracy, reduces adverse selection, and enhances profitability. It also supports the development of personalized insurance products that encourage proactive health management, representing a significant advancement in risk management within the InsurTech industry.

Rainfall projections are crucial for climate impact assessments and water resource management, especially in regions like Peninsular Malaysia (PM), where climate variability has a profound effect on socioeconomic and environmental systems. This study aims to optimize category-based statistical metrics, namely, error, efficiency, correlation, composite, and others, to reduce redundancy and identify the most reliable global climate models (GCMs) for improving rainfall projection accuracy in PM. Using CHIRPS daily rainfall data and GCM outputs from the Coupled Model Intercomparison Project Phase 6 (CMIP6), a multi-filtration approach was employed to rank GCM performance based on optimized metrics. The selected metrics span five categories, namely, 7 error metrics, 5 efficiency metrics, 3 correlation metrics, 2 composite metrics, and 5 other metrics, to address a broad range of evaluation needs, ensuring a rigorous and non-redundant assessment of GCMs. A systematic selection of GCM evaluation metrics reduced redundancy by up to 64

PurposeThis study aims to investigate the role of ta'awun ambidexterity, a new construct based on Islamic principles of collaboration, in improving business performance within Islamic organizations. It specifically examines how organizational culture, including market, clan and adhocracy cultures, influences ta'awun exploitation and exploration, and how these, in turn, affect business performance. This study introduces the concept of ta'awun ambidexterity, which uniquely integrates Islamic ethics with contemporary organizational ambidexterity theory.Design/methodology/approachA quantitative research design was used, using structural equation modeling with AMOS 24.0. This study surveyed 250 leaders of Muhammadiyah-owned enterprises (Amal Usaha Muhammadiyah [AUM]) in East Java, Indonesia. The research framework was built on organizational learning theory and integrated with Islamic business ethics. The constructs were adapted from the established literature, with ta'awun ambidexterity operationalized using newly validated Islamic business behavior indicators.FindingsThe findings show that all three cultural dimensions significantly influence ta'awun's ambidexterity. Market and clan cultures have a strong positive effect on both ta'awun exploitation and exploration. Adhocracy culture significantly affects ta'awun exploration and has a marginal effect on ta'awun exploitation. Furthermore, both ta'awun ambidexterity dimensions significantly contribute to improved business performance.Research limitations/implicationsThis study is limited to Muhammadiyah-owned enterprises in East Java, Indonesia, which may affect the generalizability of the findings to other types of Islamic organizations or regions. The study's implications suggest that spiritually grounded cooperation can enhance innovation and resource optimization in Islamic organizational settings. This study provides an empirical basis for developing strategic policies for Islamic organizations to reinforce the synergy between adaptive work cultures and Islamic values.Originality/valueThis study introduces and defines ta'awun ambidexterity as a novel construct, merging Islamic ethics with contemporary organizational theory. By embedding values such as mutual assistance, trust and shared purpose, it offers a distinct perspective on how Islamic organizations can enhance their competitiveness while adhering to their spiritual mission. This study bridges the gap between spirituality and managerial rationality in value-based organizational governance.

Indonesia is the world’s largest archipelagic state, where sea transport is the backbone of passenger and cargo mobility, yet ship accidents remain frequent and often highly fatal. This paper proposes and tests a structural model linking four human-factor dimensions—crew competence, work stress, communication effectiveness, and supervision—to perceived ship accident risk in Indonesia, with safety culture as a mediating variable. A quantitative survey is conducted among 300 seafarers operating in three major maritime regions (Batam, Merak, and Bali), and data are analyzed using Structural Equation Modeling (SEM) following recommended rules of thumb for sample size and model evaluation. The study aims to identify the most influential human-factor determinants of perceived accident risk and to provide evidence-based recommendations for strengthening maritime safety culture.