Generalized Least Square Feature Engineering-Based Weighted Gradient Boost SVM Classifier for Medical Data Diagnosis

Abstract

Generalized least square feature engineering-based weighted gradient boost SVM classification (GLSFE-WGBSC) technique is proposed. GLSFE-WGBSC technique is designed for predicting brain tumor disease with higher prediction rate and minimum time. GLSFE-WGBSC technique used generalized least square (GLS) model where feature transformation and feature creation are performed to construct new medical features from medical dataset. After creating the features, GLSFE-WGBSC technique applied Pearson chi-squared test with the aim of extracting the more significant features for disease classification with minimal time. Next, GLSFE-WGBSC technique designs a WGB-SVMC algorithm to classify the patient’s data as normal or abnormal with the support of extracted features with lower false positive rate. Thus, GLSFE-WGBSC technique achieves higher classification performance for brain tumor disease identification.