Comparison and Evaluation of Euclidean Distance and Arccosine Distance in Adaptive K-Means Clustering Algorithm for Penguin Species Clustering

Clustering is an important method in unsupervised learning for grouping data based on similarity of characteristics. This study aims to cluster penguin species based on weight, height, and wing length attributes using the K-Means algorithm with two distance approaches: Euclidean and Arccosine. The dataset consists of 342 data points after cleaning. Evaluation results show that the Arccosine distance yields a clustering accuracy of 89.6%, higher than the Euclidean distance at 63.09%. This indicates that Arccosine is more optimal for classifying penguin species.