Duration : September 2017
Supervisor: Prof. Frans Coenen
Majority of the clustering algorithms need atleast one parameter to carry out the clustering based on that parameter. This is a user specified parameter. Contrary to that, this work proposes a novel method to carry out clustering such that user does not have to provide any parameter. The algorithm carries out supervised learning on a few already clustered datasets and derives a formula which itself gives the best clustering for unknown datasets. Thus without requiring any parameter and just by analysing the dataset, the algorithm proposes the most suitable cluster for the dataset.