Optimizing Superior Seed Bunches: A New Approach to Artificial Pollination with Machine Learning Models

This study aims to optimize oil palm production through the application of more targeted pollination techniques using a data-driven approach and a new optimization model. The main focus of the study is to develop an optimization method that can improve the quality of fruit bunches and minimize waste in the production process. The research methodology begins with data collection from oil palm plantations that implement pollination under different conditions. The data is then processed through cleaning, normalization, and standardization stages to ensure consistency and reliability. The K-means clustering algorithm is applied to group data based on similarities in key variables. Furthermore, a new optimization model is designed to maximize production results based on the clusters formed, with the aim of increasing the number of good seeds, bunch weight, and minimizing rejected seeds. The results of the study showed that the application of this optimization model successfully had a positive impact on various aspects of production. The model evaluated using cross-validation showed very good performance, with a low Mean Squared Error (MSE) value (0.118 to 0.196) and an R-squared value that was close to perfect (0.998 to 0.999). This shows the model's ability to predict production results with minimal error. This optimization model improved production consistency by reducing inter-cluster variation, increasing the average number of good seeds by 13.6% and bunch weight by 13.0%. In addition, the number of rejected seeds was reduced by 28.6%, indicating better selection efficiency and improved production quality. Operational decision-making became more focused, resulting in more consistent output quality and industry standards. Resource efficiency was also improved with a 20% reduction in waste, which had a positive impact on profitability, increasing profits by 15%. Overall, the model evaluation showed an increase in quantitative output while strengthening production quality and efficiency. The novelty of the study lies in the development of an optimization model that considers multiple determinants, such as pollen viability, bunch weight, number of good seeds, and rejected seeds, to maximize oil palm production more efficiently. Keywords: Seed bunch optimization, pollination strategy, pollen viability, optimization model, machine learning, k-means clustering, multi-objective optimization.