Data-Driven Decision Support System for Sustainable Fishpong Management Using IOT and Machine Learning for Algae Bloom Prediction

Abstract

Traditional aquaculture management frequently relies on manual water quality monitoring, which often fails to detect the rapid environmental shifts preceding harmful algal blooms (HABs). This reactive approach exposes fishpond operations to significant biological risks and financial instability.

The primary goal of this research was to design and evaluate a Data-Driven Decision Support System (D3S) that integrates Internet of Things (IoT) sensors with machine learning (ML) algorithms for real-time monitoring and proactive algae bloom prediction.

Utilizing a developmental research design and the Input–Process–Output–Outcome (IPOO) framework, an IoT sensor array was deployed at a commercial fishpond in Zamboanga del Sur. The system collected high-frequency data on temperature, pH, dissolved oxygen, salinity, turbidity, and total dissolved solids. These data informed an ensemble of predictive models, including Random Forest, Gradient Boosting, and Linear Regression, to forecast chlorophyll-a concentrations.

The D3S achieved a system uptime of 99.7% and a data transmission success rate of 97.9%. Predictive analysis provided an average lead time of 4.2 hours for HAB alerts. Under the stable environmental conditions of the test site, the Linear Regression model demonstrated exceptional performance with an R² value of 0.999. User Acceptance Testing yielded mean ratings between 3.4 and 4.0, indicating high stakeholder approval.

The integration of IoT and ML effectively transitions aquaculture management from a reactive to a proactive state. By providing reliable foresight, the D3S facilitates a 20% reduction in fish mortality and enhances overall pond productivity.