Background: Banana peel waste from Ambon bananas contains essential nutrients that support the growth of Acetobacter xylinum in nata production. Differences in cultivation environments, such as highland and lowland areas, may influence the composition of banana peels and subsequently affect the characteristics of nata de banana produced.

Aims: This study aimed to analyze the differences in characteristics of nata de banana produced from Ambon banana peels cultivated in highland and lowland areas, focusing on physical properties and sensory acceptance.

Methods: An experimental design was used with banana peels from highland and lowland areas as the differentiating variables. Data on nata characteristics were analyzed statistically using an independent sample t-test with SPSS version 23.

Result: The results showed significant differences in thickness and weight of nata de banana between the two sources. Nata produced from highland banana peel extract had greater thickness and weight. Organoleptic tests also indicated that nata from highland peels was more preferred in terms of color, texture, taste, and aroma.

Conclusion: Ambon banana peels from highland areas demonstrate higher potential as raw material for nata production, offering an effective approach to utilizing local natural resources and improving the management of banana peel waste.

Background: Effective wildlife monitoring is crucial for conservation, but traditional methods are often invasive or lack spatial precision. Passive acoustic monitoring offers a non-invasive alternative, yet deriving meaningful spatial data from sound recordings remains a technical challenge, limiting its utility for detailed ecological analysis.

Aims: This study aims to design and simulate a proof-of-concept, low-cost acoustic localization system. The goal is to translate Time Difference of Arrival (TDOA) data from a simple tetrahedral microphone array into two-dimensional spatial heatmaps, providing a visual and quantitative tool to map animal vocal activity for enhanced biodiversity assessment.

Methods: A cross-shaped, four-sensor array was modeled. A custom MATLAB GUI was developed to simulate TDOA data from multiple sound sources at varied positions. The system processed this data to generate and compare four distinct types of spatial heatmaps: Gaussian Smoothing, Kernel Density Estimation, Grid Counting, and Inverse Distance Weighting

Result: The simulation successfully generated all four heatmap types, validating the core data processing pipeline. The system provided estimated source coordinates with a root mean square error (RMSE) of 0.15-0.25 meters in a controlled 6x6m area and output key statistical metrics like cluster density and distribution.

Conclusion: The prototype establishes a feasible framework for transforming raw acoustic signals into actionable spatial intelligence. This work provides a foundational step towards developing affordable, automated systems for long-term ecological monitoring, with future integration of machine learning promising direct species identification and behavioral insight.

Background: Solar energy systems are expanding rapidly, which increases the need for efficient power extraction and accurate power forecasting. Conventional maximum power point tracking methods show reduced performance under varying meteorological conditions, which leads to power losses. Machine learning offers data driven models that adapt to changing environmental patterns and improve system performance.

Aim and Scope: The study aims to enhance solar power harvesting and forecasting through machine learning techniques. Multiple predictive models are evaluated to identify reliable approaches for photovoltaic system applications.

Methodology: Solar and meteorological datasets were preprocessed through data cleaning, removal of missing values, and extraction of time based features to support time series modeling. Linear regression, random forest, and artificial neural network models were trained and evaluated through mean absolute error, root mean square error, coefficient of determination, and graphical performance analysis to achieve accurate solar power prediction and effective maximum power point tracking.

Results: The proposed framework improves solar power collection and contributes to grid stability. Machine learning based models demonstrate fast and accurate maximum power point tracking with consistent power output and improved efficiency.

Conclusion: The integration of intelligent control and machine learning techniques enhances the efficiency and reliability of solar energy systems. The proposed approach supports increased power generation, improved grid stability, and stronger sustainability of renewable energy utilization

Background: The demand for natural, environmentally friendly, and sustainable cosmetic products continues to increase along with growing consumer awareness of the negative impacts of synthetic chemicals. Coffee is one of the natural ingredients with potential application in soap formulation, as it contains bioactive compounds beneficial for skin health.

Aims: This study aimed to analyze and compare the physicochemical quality of coffee soap based on variations in manufacturing methods (cold process and hot process) and type of coffee roasting (light, medium, and dark).

Methods: The research was conducted at the Laboratory of Mathematics and Natural Sciences, University of Bengkulu, and the Laboratory of the Faculty of Agricultural Technology, Gadjah Mada University, using a two-factor Completely Randomized Design (CRD). The quality parameters observed were moisture content, pH, free fatty acids, free alkali and caffeine. Data were analyzed using Two-Way ANOVA.

Result: soap making method and the type of coffee roasting had a significant effect on the quality of water content, pH, free fatty acids, and caffeine with a significance value (p<0.05), but had no significant effect on the quality of free alkali (p>0.05) that is 0.136.

Conclusion: The combination of the cold process and medium roasting level is recommended as it provides the best balance between the physical and chemical quality of coffee soap. This study contributes to the development of coffee soap as a high-quality and competitive natural cosmetic product

Background: Biodiesel production efficiency is strongly influenced by reaction method, temperature, and reactant feeding strategy. Semi-batch transesterification offers better molar ratio control and reduced methanol waste compared to conventional batch systems.

Aims: This study analyzes the effect of feed addition frequency and reaction temperature on biodiesel efficiency and characteristics using Refined Bleached Deodorized Palm Oil (RBDPO) as raw material.

Methods: A Complete Block Design with two factors was applied: feed addition frequency (4×, 5×, 6× per period) and reaction temperature (40°C, 50°C, 60°C) with two replications. Biodiesel was produced using semi-batch transesterification with sodium methylate catalyst. Parameters measured included yield, density, pH, water content, glyceride profile, and methyl ester content. Data were analyzed using Duncan’s Multiple Range Test (5%).

Result: The best treatment was 5× feed addition, 50°C, producing the highest methyl ester content (67.18%), yield (92.98%), density (877 kg/m³), pH 6.78, and low water content (1,401 ppm). Most quality parameters approached SNI 7182:2015 biodiesel standards.

Conclusion: Semi-batch operation improves conversion control but is not yet fully optimal due to reverse reactions. Further optimization of methanol ratio and reaction time is required to suppress monoglyceride and diglyceride reformation.