Predicting Abundance of Fish Species Populations in Manila Bay, Philippines Based on Ensemble Learning Approach

Manila Bay is a significant contributor to the Philippines' fish production, but its resources have been depleted due to overfishing, pollution, and damage, leading to a decline in fish catch and a shift towards less valuable species. Conventional approaches to fisheries stock assessment impose constraints on our comprehension of fish population dynamics. These limitations can be overcome through the utilization of machine learning techniques, which enable the forecasting and modeling of fisheries populations with improved accuracy and understanding. In this study, the commercial fisheries populations data collected from 2018 to 2021 in Manila Bay were used to predict the abundance of species fisheries production data using the K-NN - MLP - Logistic Regression (KNMLPR) model based on the majority voting ensemble approach. Analysis revealed that it is possible to combine the strengths of multiple models and improve overall predictive performance. The results also suggest that the k-nearest neighbors and logistic regression models have the best performance in predicting fish species population dynamics, while the neural network model shows slightly lower accuracy. This study provides valuable insights for fishery management and policymaking to support sustainable fishing practices in the region. Further research could focus on exploring additional machine learning algorithms and incorporating environmental factors to improve the prediction accuracy of the model.