A New Visualization of Group-Outliers in Unsupervised Learning

Abstract

This paper presents a new method for computing a quantitative score which can help in detecting cluster outliers using visualisation task. Self-organising map is incorporated in the proposed approach. The proposed method is evaluated on a number of datasets from UCI. Visualizations and experimental results show that GOF sensibly improves the results in term of cluster-outlier detection. The development of the SOM based visualization tool intends to provide additional exploratory data analysis techniques by offering a tool that allows effective extraction and exploration of patterns.