期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2014
卷号：67
期号：3
出版社：Journal of Theoretical and Applied
摘要：Existing Data Mining techniques, concentrates mainly on finding patterns in large datasets and it also focuses on concepts such as classification, association rules and clustering. Very few work aims on finding Boundary values in classification. Boundary values are the small subset of the data set which contains critical information useful for predicting accurate class labels for the new instances. This paper describes technique for detecting boundary records using different distance based approach during classification task and computes the computational time for extracting the instances using different distance metrics. The records isolated as boundary region of the specified class, can contain domain-specific information which is useful for finding critical records to improve the classification accuracy. This information is useful for further decision making in classification. Experiments are carried out for numeric data set and results shows that only subset of the data set are found to be boundary set and compares the computational time for extracting the boundary region using the distance metrics.
