Kernelized Normal Discriminant Feature Selection and Borda Count Bootstrap Aggregating Classification for Risk Factor Identification and Disease Diagnosis
P. S. Renjeni1, B. Mukunthan2, G. Rakesh3
1P. S. Renjeni*, Research Scholar, Jairams Arts & Science College, (Affiliated to Bharathidhasan University) Karur.
2B. Mukunthan, Research Supervisor, Jairams Arts & Science College, (Affiliated to Bharathidhasan University) Karur.
3G. Rakesh, Dean of Science, Department of Computer Science, Jairams Arts and Science College, (Affiliated to Bharathidhasan University) Karur – 639003, Tamilnadu, India.
Manuscript received on April 11, 2020. | Revised Manuscript received on April 20, 2020. | Manuscript published on April 30, 2020. | PP: 2389-2395 | Volume-9 Issue-4, April 2020. | Retrieval Number: C6334029320/2020©BEIESP | DOI: 10.35940/ijeat.C6334.049420
Open Access | Ethics and Policies | Cite | Mendeley
© 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: : Automatic detection of disease is crucial in health care management to evaluate large patient data. The early diagnosis and treatment of disease are an important task to prevent the patient from death. The various researchers have contributed to the development of disease diagnosis. But still, it causes the more risk for identifying the patient health conditions. In order to improve the disease diagnosing accuracy, A Kernelized Normal discriminant Feature Selection based Borda count bootstrap aggregating Classification (KNDFS-BCBAC) technique is introduced for identifying the patient health condition and critical factor analysis with higher accuracy and lesser time. At first, radial basis kernelized normal discriminant analysis is used to identify the relevant feature for minimizing the complexity of disease diagnosis. After selecting the relevant features, Borda count bootstrap aggregating Classifier is applied to classify the patient data as abnormal or normal by constructing the weak learner as bivariate correlated regression tree. Then, the diseased data is considered as a training sample for analyzing the critical factor and classifies the patient data level as the initial stage, critical stage based on the threshold range of features value. By applying the Borda count voting scheme, the weak learner results are combined into strong. In this way, disease diagnosis and critical factor analysis of patient data are performed with greater accuracy and minimal time complexity (TC). Experimental is performed with tumor dataset on metrics namely disease diagnosis accuracy (DDA), false alarm rate (FAR), and TC. The observed results evident that KNDFSBCBAC technique achieves higher DDA with lesser complexity and FAR than the conventional methods.
Keywords: Disease diagnosis, feature selection, radial basis kernelized normal discriminant analysis, bootstrap aggregating Classification, bivariate correlated regression tree, Borda count voting scheme.