A New Classwise k Nearest Neighbor (CKNN) Method for the Classification of Diabetes Dataset
Y. Angeline Christobel1, P. Sivaprakasam2
1Y. Angeline Christobel, Research Scholar, Department of Computer Science, Karpagam University, Coimbatore, India.
2Dr. P. Sivaprakasam, Professor, Department of Computer Science, Sri Vasavi College, Erode, India.
Manuscript received on January 20, 2013. | Revised Manuscript received on February 15, 2013. | Manuscript published on February 28, 2013. | PP: 396-400 | Volume-2 Issue-3, February 2013.  | Retrieval Number: C1155022313/2013©BEIESP

Open Access | Ethics and Policies | Cite
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: The general problem for data quality is missing data. The real datasets have lot of missing values. Mean method of imputation is the most common method to replace the missing values. In our previous work [23], we address the negative impact of missing value imputation and solution for improvement while evaluating the performance of  algorithm for classification of Diabetes data. In this paper, we address a new Class-wise k Nearest Neighbor  method for the Classification of Diabetes Dataset. We selected diabetes dataset because it contains lot of missing values and the impact of imputation is very obvious. To measure the performance, we used Accuracy, Sensitivity and Specificity and Error rate as the metrics. The arrived results show the significant improvement measured with respect to the above metrics.
Keywords: Data Mining, Classification, kNN, Imputation, Data Normalization and Scaling.