Predictive Modeling of pH in a Small-Scale Aquaponics System: Multi-Layer Perceptron (MLP) Regression, Support Vector Regression (SVR) and Random Forest Regression Models | ||||
| Egyptian Journal of Aquatic Biology and Fisheries | ||||
| Article 5, Volume 29, Issue 1, January and February 2025, Page 101-115 PDF (1.08 MB) | ||||
| DOI: 10.21608/ejabf.2025.403145 | ||||
 View on SCiNiTO | ||||
| Author | ||||
| Rharrhour et al. | ||||
| Abstract | ||||
| Aquaponics is a growing industry that combines intensive food production with waste-stream recycling and water conservation, offering alternative solutions to soil degradation and water scarcity. This technique can contribute to global food security but requires careful management. One of the key parameters in aquaponics is pH, which must be maintained to accommodate three different types of living organisms: fish, plants, and bacteria. In aquaponics systems, pH naturally decreases due to the nitrification process, making monitoring essential. To predict pH levels in a small-scale aquaponics system—consisting of three hydroponic techniques (DWC, media bed culture, and NFT) combined with a tilapia fish tank—three machine learning models were proposed in this study. The results showed that the random forest regressor model can predict pH fluctuations over 12 days with a root mean square error (RMSE) of 0.0260 and a mean squared error (MSE) of 0.0006. The random forest model outperformed the MLP regressor and SVR models in terms of accuracy, suitability, and prediction error. Predicting pH is crucial for the stability of an aquaponics system. | ||||
| Keywords | ||||
| Aquaponics; Biodiversity; Food security; Machine learning; MLP Regressor; SVR; RF Regressor | ||||
|
Statistics Article View: 295 PDF Download: 353 |
||||
