Clustering is a widely used technique in data analysis, but outliers can significantly impact the accuracy of traditional clustering methods. To address this issue, a new clustering technique is proposed that incorporates outliers during clustering. The proposed approach involves using a variable, (λ > 0), to define the cluster radius. Weighted and Unweighted clustering methods are proposed for clustering considering the outliers. Centroids are calculated without using the direct K-means algorithm. Evaluation on standard datasets, such as Fisher Iris, shows improved performance compared to K-means and DBSCAN, with a maximum Silhouette score of 0.568 and an accuracy of 88% even in the presence of outliers.
-
Notifications
You must be signed in to change notification settings - Fork 0
TejaswiniKatale/New-Approaches-to-Robust-Homogeneous-And-Clearly-Identifiable-Cluster-Creation
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
About
A new clustering technique is proposed that incorporates outliers during clustering. The proposed approach involves using a variable, (λ > 0), to define the cluster radius. Weighted an
Topics
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published