Абстрактный

Clustering of Data with Mixed Attributes based on Unified Similarity Metric

M.Soundaryadevi, Dr.L.S.Jayashree

Most of the clustering approaches are applicable to purely numerical data or purely categorical data but not both. There exists an awkward gap between the similarity metrics for categorical and numerical data, so it is a non trivial task for clustering of data with mixed attributes. A general clustering algorithm for based on object cluster similarity is framed which clusters the data with mixed attributes. Moreover, clustering techniques are applied in Educational Data Mining (EDM) to group of students according to their customized features. Student data set is a collection of students’ personal characteristics, skill profiles, etc. It mostly contains the collection of attributes of different types. So, computationally efficient and simple clustering algorithm is required for this kind of data sets. The group of students obtained from clustering technique can be used by the instructor to build an effective learning system, to promote group learning, to provide adaptive contents etc. Here an iterative clustering algorithm based on object cluster similarity (OCIL) is to be applied on the student data set with mixed attributes. This will serve as a valid guide line for development of intelligent tutoring system.