Invasive Weed Optimization K-Means Performance Robust Operations (IWOKM PRO) in High-Dimensional Datasets
Ni Luh Gede Pivin Suwirmayanti, I Ketut Gede Darma Putra, Made Sudarma, I Made Sukarsa, Emy Setyaningsih, Ricky Aurelius Nurtanto Diaz · Engineering Technology & Applied Science Research · 2025
This study presents a novel clustering approach called Invasive Weed Optimization K-Means Performance Robust Operations (IWOKM PRO) to improve clustering performance on high-dimensional datasets. Unlike previous IWOKM implementations, IWOKM PRO focuses on optimizing parameter efficiency to conserve computational resources and applies centroid selection techniques to accelerate convergence and enhance clustering results. To evaluate its effectiveness, IWOKM PRO was tested on stock data collected from the Indonesia Stock Exchange (IDX), comprising 604 stocks with adjusted closing price features from January 2019 to December 2023. The experimental results demonstrate that IWOKM PRO outperforms the original IWOKM method in both convergence speed and clustering accuracy. Specifically, in the three-cluster scenario, IWOKM PRO achieved the best fitness value in 1.37 s with a Sum of Squared Errors (SSE) of 973.6434. In the five-cluster scenario, IWOKM PRO reached an average convergence time of 6.45 s with an SSE of 443.8437. Compared to IWOKM, these results significantly improve computational efficiency and clustering performance. In general, this study shows that IWOKM PRO is an effective solution to improve the efficiency and accuracy of clustering, particularly for high-dimensional financial datasets.