Machine Learning-based Assessment of Automated Composting System for Banana Agricultural Waste

The demand for bananas has been growing over the decades while also generating considerable amount of waste. Such agricultural waste, if no proper waste management, can have detrimental environmental effects. This study presents an automated composting system for efficient waste management to improve compost quality and maturity of banana agricultural waste. It outlines the system's design which monitors the temperature, humidity, pH, and moisture of the compost mixture and automates the control for mixing, moistening, ventilation, and heating. A comparative assessment involving two methods of traditional composting, the open-air composting and composting bin method as benchmarks, and the automated composting system. Data Analysis techniques such as Principal Component Analysis (PCA), Decision Tree Analysis, and Two Sample t-tests, alongside ensemble learning methods evaluated the variable’s effects on the composting process and the compost maturity recognition. The results show that the automated composting system produced quality and mature compost within 10 days, outperforming benchmark results by up to 2.25 times faster by maintaining crucial variable values.