<![CDATA[Abstrak. Penurunan kualitas sumber air menjadi permasalahan penting yang memerlukan metode deteksi yang cepat, murah, dan mudah diimplementasikan. Pemantauan kualitas air secara konvensional umumnya dilakukan melalui analisis laboratorium yang membutuhkan biaya tinggi serta waktu yang relatif lama. Penelitian ini mengimplementasikan machine learning untuk mengklasifikasikan kualitas air berdasarkan analisis fitur warna digital dalam ruang warna RGB dan HSV. Dataset yang digunakan terdiri dari 647 citra air yang diklasifikasikan ke dalam dua kelas, yaitu air jernih dan air kotor. Fitur warna diekstraksi berupa nilai rata-rata R, G, B, H, S, dan V, kemudian dilakukan pelabelan otomatis berdasarkan threshold pada ruang warna HSV. Beberapa algoritma machine learning, yaitu Random Forest, Support Vector Machine, K-Nearest Neighbor, dan Decision Tree, digunakan dan dibandingkan kinerjanya. Hasil penelitian menunjukkan bahwa Decision Tree menghasilkan akurasi tertinggi sebesar 100%, diikuti oleh Random Forest sebesar 99,23%, Support Vector Machine sebesar 98,46%, dan K-Nearest Neighbor sebesar 96,92%. Analisis feature importance menunjukkan bahwa komponen Value dan Saturation pada ruang warna HSV merupakan fitur paling dominan dalam proses klasifikasi. Hasil ini menunjukkan bahwa kombinasi machine learning dan fitur warna digital efektif digunakan untuk klasifikasi kualitas air secara otomatis. Abstract. The declining quality of water resources requires fast, cost-effective, and easily deployable detection methods. Conventional water quality monitoring relies on laboratory analysis, which is expensive and time-consuming. This study implements machine learning to classify water quality based on digital color feature analysis in RGB and HSV color spaces. The dataset consists of 647 water images classified into clear and turbid water categories. Color features were extracted as mean values of R, G, B, H, S, and V, followed by automatic labeling using HSV-based threshold rules. Several machine learning algorithms, including Random Forest, Support Vector Machine, K-Nearest Neighbor, and Decision Tree, were evaluated and compared. The results indicate that the Decision Tree achieved the highest accuracy of 100%, followed by Random Forest at 99.23%, Support Vector Machine at 98.46%, and K-Nearest Neighbor at 96.92%. Feature importance analysis reveals that Value and Saturation components in the HSV color space are the most influential features for classification. These findings demonstrate that machine learning combined with digital color features provides an effective approach for automated water quality classification.]]>
