Enhancing clustering performance: an analysis of the clustering based on arithmetic optimization algorithm

This study explored the clustering based on arithmetic optimization algorithm (CAOA) and its potential for addressing challenging clustering problems. CAOA is based on the arithmetic optimization algorithm (AOA), which utilizes arithmetic operators, including Addition, Subtraction, Multiplication, and Division, to optimize solutions. The performance of CAOA was investigated by applying it to diverse real-life datasets and meticulously analysing its clustering performance. Two primary evaluation metrics, namely the average distance among cluster members (intra-cluster distance) and the F-measure, were employed to gauge the clustering quality. Statistical validation was conducted using the Friedman test, ensuring robust and significant results. The results revealed substantial insights into CAOA's performance. In terms of average intra-cluster distance, CAOA consistently recorded the lowest values among all tested clustering algorithms. This outcome indicated CAOA's ability to form tightly packed, well-defined clusters, enhancing its suitability for applications like pattern recognition and data segmentation. Regarding F-measure, CAOA delivered competitive clustering quality. Notably, it achieved among the highest F-measure values, especially in datasets like "Cancer" and "LR," signifying its potential for accurate cluster identification, crucial in domains such as medical diagnosis and customer segmentation. This study indicated the effectiveness of CAOA in addressing real-world clustering challenges. The findings emphasized CAOA's consistent superiority over other algorithms in minimizing the average intra-cluster distance while also demonstrating competitive clustering quality as measured by the F-measure. Statistical validation through the Friedman test confirmed the distinctiveness of CAOA's performance.