Automatic Clustering by Geometric Method of Tangent Lines with Points

Clustering is the division of data samples into separate groups based on standard features to discover their structural properties. In recent studies, researchers introduced the peak clustering algorithm, which combines density-based and center-based clustering. However, the algorithm has two main weaknesses. (1) Insufficient accuracy in determining cluster centers in low-density clusters (2) Recognizing actual data points as noise data points. Researchers presented the Extreme clustering algorithm as a more efficient solution that overcomes the weaknesses of the Peak Clustering algorithm while still benefiting from its advantages. However, the Extreme clustering algorithm is not very accurate in separating the boundaries of overlapping clusters. We have developed an automatic clustering solution based on the geometric method of tangent lines with points. Our method uses the Extreme clustering algorithm to determine the Extreme points. Then, it applies the Apollonius problem and its definition of polygons to cover all the points until it reaches the exact clustering boundary. After evaluating the results, the proposed clustering algorithm in this paper can increase the accuracy of separating the boundaries between different clusters.