GENERIC FRAMEWORKS FOR SVM, ANN, LGBM, AND LR ALGORITHMS
Nora Ibrahem Alghurair; Mohammad A. Mezher
Journal Title:International Journal of Computer Science and Mobile Computing - IJCSMC
World Health Organization describes diabetes as a multiple etiological metabolic condition defined by persistent hyperglycemia with anomalies in glucose, lipid and protein metabolism triggered by insulin secretion deficiencies, lipid or both. Diabetes is one of the 21st century's most daunting health problems in the world, and affects over 425 million people. Data mining is one of the major techniques which develops and supports medical data research. The aim of this research was to establish a diagnostic framework for diabetes. The PIDD used to function description. Two generic frameworks have been proposed in the study. The first framework uses the ANN technique and fed production to SVM which yields the diagnostic result. With this generic framework, five experiments are carried out and the highest accuracy achieved was 81.8%. The second framework employs an ensemble of majority voting techniques which combines LGBM, SVM, and LR. The generic framework was right at 87.9 per cent. The frameworks presented in contrast with other state-of-the-art solutions, and it found that the second solution is the better one.