PERFORMANCE VALIDATION OF PRIOR QUANTIZATION TECHNIQUES IN OUTLIERS CLASSIFICATION USING WDBC DATASET
Journal: INTERNATIONAL JOURNAL OF ENGINEERING TECHNOLOGIES AND MANAGEMENT RESEARCH (Vol.5, No. 4)Publication Date: 2018-04-30
Authors : Dr.D.Rajakumari;
Page : 48-56
Keywords : Data Mining; Classification; Outlier Detection; Feature Selection; Wisconsin Diagnosis Breast Cancer (WDBC).;
Abstract
Data mining is the process of analyzing enormous data and summarizing it into the useful knowledge discovery and the task of data mining approaches is growing quickly, particularly classification techniques very efficient, way to classifying the data, which is important in the decision-making process for medical practitioners. This study presents the quantization and validation (OQV) techniques for fast outlier detection in large size WDBC data sets. The distance metrics utilization makes the algorithm as the linear one for various objects and assures the sequential scanning. The inclusion of direct quantization technique and the cluster explicit discovery assures the simplicity and the economical. The comparative analysis of proposed OQV techniques with the triangular boundary-based classification and the Weighing-based Feature Selection and Monotonic Classification (WFSMC) regarding the accuracy, precision, recall and the number of attributes assures an effectiveness of OQV for large size datasets.
Other Latest Articles
- A STUDY REGARDING AWARENESS ABOUT WEANING AMONG MOTHERS VISITING PEDIATRIC MEDICINE DEPARTMENT
- DYNAMICS OF ENVIRONMENTAL POLLUTIONTHROUGH VEHICLES
- EVALUATION OF FUNCTIONAL OUTCOME OF HIP FRACTURE SURGERY IN ELDER PATIENTS
- Maritime English Teaching Instruction In Global Maritime Context
- INFLUENCE OF PHOSPHORUS IMPURITY ON THE STRUCTURE AND NATURE OF THE DESTRUCTION OF THE GENUINE DAMASCUS STEEL
Last modified: 2018-05-21 17:13:58