<![CDATA[The continuous increase in daily waste accumulation has become a major issue in many areas, primarily due to the mixing of various waste types and the lack of effective household waste management. This complicates waste processing and contributes to environmental degradation. This study aims to design and implement a practical tool for detecting organic and inorganic waste types, specifically for use by household waste collection personnel. The developed system utilizes three sensors, capacitive, inductive, and light-dependent resistors (LDR), to acquire characteristic data from different types of waste. The device is designed in the shape of a pistol to enhance mobility and ease of use by waste collection officers. For the waste-type classification system, several machine learning methods were employed, namely Adaptive Boosting (AdaBoost), Support Vector Machine (SVM), and K-Nearest Neighbor (KNN). Based on the experimental results, AdaBoost was selected as the primary model for the waste classification system because of its superior performance in terms of cross-validation accuracy and the balance of evaluation metrics, such as precision, recall, and F1-score. Consequently, AdaBoost predictions were adopted to establish a rule-based classification logic by extracting threshold values from the most influential sensor features. This study utilized AdaBoost analysis as the foundation for rule formulation, ensuring that classification decisions were based on reliable and tested data patterns. Based on testing with several samples, the device can classify organic and inorganic waste types with an accuracy rate of 91.67%. Additionally, the tool can estimate the composition of mixed waste with an error rate of 5.06%. The presence of this device has been proven to accelerate and simplify the waste-sorting process, thereby increasing the efficiency of household waste management.]]>
