Clustering Process for Mixed Dataset Using Shortest Path Non Parameterised Technique

Abstract

Clustering in mixed dataset is a dynamic research focus in data mining concepts. The predictable clustering algorithm related to be more supportive to only one kind of attribute not for the mixed data type. Hence, the traditional clustering techniques processed with mixed attributes either by converting the numerical data type to categorical type or categorical type to numerical data type. But, utmost of the clustering processes are improved by converting numerical attributes. This progression of grouping ends up with two boundaries, the earlier limitation is that conveying numerical values to all types of categorical data is simply difficult. On the other hand the later drawback lies in the parameterized clustering which needs number of clusters as response for grouping the datasets. To succeed over the limitations the clustering technique is organised by incorporating shortest path and non-parameterized clustering. The proposed work of Shortest path non parameterised Clustering technique, the input parameter (number of clusters) is discovered spontaneously and the data objects of the cluster are grouped that are at the shortest distance.