Enhancing K-means clustering accuracy based on robust and abs scaling

We discuss a new narrative custom scaling method employed improving to improve the outcome of K-Means Clustering in the current article. The new method, referred to as custom scaling, is defined as X′= X−Median(X) / max(X), a combination of the robustness of median-based scaling and the normalization provided by max-based scaling. This approach effectively reduces the influence of outliers and ensures that feature values are appropriately normalized. The comparison of K-means clustering outcome is often made with traditional scaling methods (standard scaling and min-max scaling) and our custom scaling technique across several datasets. Experiments reveal the facts on how the proposed custom scaling method significantly improves the accuracy of clustering, especially for datasets with outliers.