Determining appropriate of Data Classification with Multi-Layer Perceptron, Support Vector Machines and Radial Basis Function

Abstract

This research is using data mining to compare the identification information of the model created by the Multi-layer Perceptron Neural Network, Support Vector Machines and Redial Basis Function. The model consisted of 30 models of three series data from the UCI that features different is Vote, Audiology and Ionosphere to measure the efficiency and accuracy of data classification, Precision, Recall and quality measurement (F-measure). The results showed that the model based on support vector machine (SVM) performance is 97.25%, followed by the creation of a Redial Basis Function (RBF) is 96.80% and the final model based on Multilayer Perceptron (MLP) is 96.09%, respectively.