Probit Regressed Feature Selection Based Linear Programming Boost Classification for Tumor Risk Factor Identification and Disease Diagnosis
P. S. Renjeni1, B. Mukunthan2
1P. S. Renjeni, Research Scholar, Jairams Arts & Science College, Karur-3.
2Dr. B. Mukunthan, Research Supervisor, Jairams Arts & Science College, Karur-3.
Manuscript received on 01 August 2019. | Revised Manuscript received on 09 August 2019. | Manuscript published on 30 September 2019. | PP: 8152-8160 | Volume-8 Issue-3 September 2019 | Retrieval Number: C6087098319/2019©BEIESP | DOI: 10.35940/ijrte.C6087.098319
Open Access | Ethics and 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: Accurate diagnosis of survival rate in patients with tumor remains challenges due to the increasing complexity of treatment protocols, and different patient population samples. Due to the complexity, the risk factor of the patients gets increased. Therefore, a reliable and well-validated prediction needs to develop the automatic disease diagnosis for early detection of the tumor. The novel technique called Probit Regressed Feature Selection based Iterative Linear Programming Boost Classification (PRFS-ILPBC) is introduced for tumor risk factor identification and disease diagnosis of patient data with higher accuracy and lesser time consumption. In PRFS-ILPBC technique, Probit Regression model is a regression type to estimate the relationship between the features and the disease symptoms using bivariate correlation coefficient. Based on the correlation results, the features fall into any one of the two classes (i.e. relevant or irrelevant). With the help of relevant feature, Iterative Linear Programming (LP) Boost Classification model is applied to perform classification by combining the weak learner for tumor risk factor identification and disease diagnosis. LPBoost constructs the strong classifier through initiating with a set of weak classifiers. The training data (i.e. patient data) are taken as the input and added to the set of considered weak classifiers. The kernelized support vector machine act as weak learner compares the training features with the testing results to identify the risk factor and classify the patient data into normal or abnormal. The ensemble classifier improves disease diagnosis accuracy and reduces the false positive rate. Experimental evaluation of proposed PRFS-ILPBC technique and existing methods are carried out with different factors such as disease diagnosing accuracy, false positive rate, and computation time with respect to a number of patient data. The observed results reported that the proposed PRFS-ILPBC technique achieves higher disease diagnosing accuracy with minimum computation time as well as false positive rate than the conventional techniques.
Keywords: Tumor Disease Diagnosis, Feature Selection, Probit Regression, Bivariate Correlation Coefficient, Iterative Linear Programming Boost Classification, Kernelized Support Vector Machine
Scope of the Article: Classification