Visual clustering through weight entropy

Cluster visualisation is an essential part in data mining to validate and refine the clusters necessarily. While much visualisation which is proposed in recent years, help the users to explore clusters and refine it necessarily. This requires an efficient and flexible human-computer interaction, which can be achieved by domain knowledge. In this paper, an integrated visual framework is proposed for cluster visualisation and validation which utilises the power of existing visual clustering model by incorporating domain knowledge through weight entropy of soft subspace clustering scenario. The efficiency of the proposed work can be analysed with the well known centroid-based partitional clustering algorithms. Experiments demonstrate that the proposed method works well with large number of dimensions and eases the human-computer interaction in an effective way. The experiments are carried out for various datasets of UCI machine learning data repository.