Identification of Breast Cancer Using Ensemble of Support Vector Machine and Decision Tree with Reduced Feature Subset
H.S Hota
H.S Hota, Guru Ghasidas Central University, (C.G),India.
Manuscript received on 13 February 2014 | Revised Manuscript received on 20 February 2014 | Manuscript Published on 28 February 2014 | PP: 99-102 | Volume-3 Issue-9, February 2014 | Retrieval Number: I1504023914/14©BEIESP
Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© 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: Breast cancer is very common disease found in woman in which breast masses are increases abnormally .A recent survey in united kingdom proved that breast cancer is not only a problem of young woman but it is also a problem of old age woman those who have crossed the age of sixty or even seventy. An early identification and then prevention with proper medication of breast cancer can save life of human being. A robust and efficient breast cancer identification system is necessary for this purpose. Statistical technique like support vector machine and data mining technique like decision tree are widely used by the researcher since last few years. These techniques proved their ability to efficiently diagnose breast cancer problem. In this research work an ensemble model based on above two techniques are explored with special reference to feature selection. A rank based feature selection technique reduces features one by one based on its rank of breast cancer data ,downloaded from UCI repository site. An ensemble of support vector machine and C5.0 decision tree technique with reduced subset of only five features produced high accuracy of 92.59%.
Keywords: Decision Tree (DT), C5.0, Support Vector Machine (SVM),Feature Selection (FS).
Scope of the Article: Machine Design