Predictive analysis using resilient and traditional backpropagation algorithm

Artificial neural networks can be used in many applications like analysis, manipulation, predictions on the given statistical data. Neuralnet is the function which is used to train the multi-layered perceptron when discussed about the regression analysis ,i.e to maximize the func-tional relationship between predictors(input variables) and the response variables(output). Hence neural networks can be used as an exten-sion of generalized linear models(supervised learning). This paper deals with the brief introduction about multi-layered perceptron and the validation of resilient backpropagation and the traditional backpropagation algorithms. A real time bank dataset is considered and predictions is being done on the Churning rates of Customers by the above two algorithms. It is observed that resilient backpropagation algorithm gives more accurate and precise results in comparison with traditional backpropagation algorithm by the number of steps taken by both the algorithms in convergence of the whole neural network and the error rate of both the algorithms.