Implementasi Algoritma Gray Level Co-Occurrence Matrix dan K-Nearest Neighbor dalam Analisis Penyakit Kulit

Abstract

Skin disease is a health disorder that can be caused by infection, inflammation, or environmental factors and requires early detection to prevent more serious impacts. This research aims to implement the Gray Level Co Occurrence Matrix (GLCM) and K-Nearest Neighbor (K-NN) methods in skin disease classification based on digital images. The data used was obtained from Kaggle, consisting of 200 images covering four types of skin diseases, namely chickenpox, herpes, ticks, and skin cancer, with a proportion of 80% for training data and 20% for testing data. The research stages include data collection, pre processing (cropping and resizing) feature extraction with GLCM, and classification using K-NN. The extracted features include contrast, energy, correlation, and homogeneity, which are then used as the basis for classification using the K-NN algorithm with various K values. Testing was carried out with several variations of the K value to determine the best level of accuracy in distinguishing skin disease types based on their texture characteristics. The results showed that the GLCM method can extract the texture features well, while the K NN algorithm achieved the highest accuracy of 100% at K=1 and K=2, but decreased at higher K values with an average accuracy of 89.50%. These results show that the methods used can help the process of early detection of skin diseases more quickly and accurately in the system.