Predicting Catfish Growth and Feed Efficiency in Using Decision Tree and Support Vector Regression

Abstract

Catfish farming has a key part in maintaining the economy of Poris Plawad Utara, Cipondoh, Tangerang where many farmers depend on it as their primary source of income. However, poor feed management creates considerable obstacles as overfeeding leads to higher expences and enviromental issues while underfeeding inhibits fish growth. Traditional methods for identifiying ideal feed amounts rely on manual observation, which often leads in irregular growth rates and feed loss. Despite the necessity of effective feed utilization, there is a paucity of powerful predictive techniques available to enable farmers accurately forecast feed demands and fish growth. There, we employ machine learning approaches including Decision Tree and Support Vector Regression (SVR), to predict catfish development and feed efficiency based on several environmental parameters such as water temperature, pH levels, and oxygen concentration. The algorithm we used was trained using data acquired from catfish farm in Poris Plawad Utara, comprising 3 month of feeding and growth records. The results of the analysis demonstrate that while Support Vector Regression (SVR) and Decision Trees perform well in stable environments, they have trouble handling environmental changes. Accuracy is impacted by feed management and environmental stability. More variables and an intricate machine learning strategy are required for better performance. While SVR works well in stable systems, complicated dynamics require adaptations. These results show that feed efficiency and fish development may be grately increased by incorporating machine learning into catfish farming operations. This methodology provides farmers with data-driven solutions that maximizes the efficiency of aquaculture and sustainability.