A Hybrid Classification Model of Artificial Neural Network and Non Linear Kernel Support Vector Machine

Machine Learning Algorithms are employed in characterization, pattern recognition, and prediction. A hybrid model helps reducing the computational complexity, improves accuracy, results an effective method for classification. The misclassification of individual classifier is often excluded a classifier. objective this research was to develop classification Artificial Neural Network non-linear kernel Support Vector as intelligent tool achieving better performance minimizing error rates. This study further evaluated irreducibility identifiability statistical properties ANN-SVM model. To achieve hybridization ANN SVM, first obtained weights from fitted model, these were used initial structure. experiment carried out three distinct phases: selection input features using Boruta Wrapper Algorithm, learning, combined effect optimization. findings suggest that approach gives higher accuracy 89.7% more precise compared single ANN, SVM data mining algorithms. Therefore, best binary system classifying diabetes mellitus. software analysis R.