Designing and Developing novel methods for Enhancing the Accuracy of Water Quality Prediction for Aquaponic Farming

Abstract

The Internet of Things (IoT)-based automated water surveillance system is crucial for fish farming, as it helps control risks and improve output and productivity. However, the industry faces challenges such as a lack of understanding of organism selection based on water purity indicators, a shortage of premium seeds and species, and imbalanced datasets. Existing systems also lack optimal feature selection methods, as well as modern machine learning and deep learning approaches. To address these issues, a novel system is proposed that integrates an aquaponic ecosystem containing fish, plants, and bacteria to balance water quality parameters and boost productivity. The system collects data from IoT devices, performs a data cleaning process using missing values, and outliers handling. Then, it performs a feature extraction process to select the optimum features. Next, it employs the novel H-SMOTE approach to tackle the issue of imbalanced datasets. The system uses multi-model categorization to cultivate fish in cold waters, warm-water aquatic plants, and bacteria in an aquaponic environment. The system uses a voting principle to identify the most effective prediction model. The proposed system achieves 99.50% accuracy for water quality prediction for aquaponic farming, addressing the limitations of existing systems and improving prediction accuracy and overall aquaponic farming productivity.