Enhanced K-Means Clustering Approach for Diagnosis Types of Acne

Abstract

Acne is a skin disorder all humans almost have, both women and men. How to treat acne properly determines how quick you will be acne-free. Still, the dependency on doctors in conducting skin physical examinations to make an early diagnosis remains high. Therefore, this research was conducted by developing K-Means Clustering model for early diagnosis of types of acne experienced by the patients. The K-Means clustering algorithm were as follows: (a) determining the total clusters; (b) allocating the data into groups, randomly; (c) calculating the centroid in each cluster; (d) allocating each data to the centroid (e) repeating the centroid calculation if there were still data moving fro one cluster to another. The results of the performance of the K-means model would produce types of acne with four output categories according to the severity of acne such as: no acne (16.12%), mild acne (29.03%), moderate acne (32.25%), and severe acne (22.60%).