Implementasi Machine Learning Menggunakan Algoritma Klasifikasi untuk Mendeteksi Jenis Sampah

Abstract

One important aspect of waste management is the grouping and sorting of waste based on its type. However, waste sorting carried out by the public is often inaccurate or inconsistent. This can be caused by a lack of knowledge about waste types, confusion in identifying the correct type, or difficulty in memorizing complex sorting guidelines. Therefore, a system is needed that can detect waste types quickly and accurately without involving a large amount of human labor. One technological solution that can be used is machine learning. The method used in this study is the Random Forest algorithm. The data used consists of waste grouped based on characteristics such as texture and shape. This data is processed through feature extraction and preprocessing before being applied to the Random Forest model. The model's accuracy is tested using cross-validation techniques to assess classification performance. The experimental results show that the Random Forest algorithm can achieve a high level of accuracy in detecting waste types. The accuracy obtained reaches 94%, with consistent results in each cross-validation fold. This model proves to be effective in classifying different types of waste using the available features. The implementation of the Random Forest algorithm in waste type detection demonstrates great potential in improving technology-based waste management systems. With high accuracy, this model can be applied to various waste classification systems in society, helping to improve the efficiency of recycling processes and waste reduction.