Open Access Open Access  Restricted Access Subscription or Fee Access

Data Mining Using Efficient Artificial Neural Network Back Propagation Algorithm

A.S. Kumaresan, Dr.E. Kannan


The Abstract uses a tri-layer feed forward Artificial Neural Network (ANN) using back propagation algorithm as a framework to differentiate customers of a German automobile company into various categories. This classification is performed on the three divisions (namely Germany, South Africa and Maldives) of the company. The breakdown of customers into different categories (namely Good, Average and Below Average.) is primarily done on the basis of invoicing data and forms an important component in the concept of Data Mining. The concept of classification in Data Mining using neural networks involves taking day to day invoicing data of the customers as the base. From this raw data the intelligent data is extracted through the process of data cleaning and relevance analysis. The data extraction is made on a number of reasons like the value of invoices for different customers, the quantity ordered through invoices by different customers and the number of invoices ordered over a period of time. This intelligent data is then conditioned averaged, prepared and normalized. Normalization makes the data suitable for a tri layer feed forward ANN using back propagation algorithm. On the basis of a number of iterations of the “supervised” Input / Output training pairs the ANN teaches to master the classification of customer data. The error in each such iteration on the ANN is fed back to adjust the weights in the last layers, there by making the network an exact classifier. The ANN uses various learning rate annealing schedules, different number of nodes in the hidden layer and different activation functions which not only provides for the study of different rates of error convergence but also the evaluate the strength and support in Data Mining. This framework permits us to predict a customer bifurcate for a new customer of the company. Moreover, this framework is even independent of customer data and would equally uses to other aspects of retail and wholesale business.


Data Cleaning, Analysis, Normalization, Neural Network, Back Propagation Algorithm, Sensitivity

Full Text:



Baldi, Hornik, Problem of Learning in Multilayer Feed Forward Neural Network, 1989, pp. 3-5.

Cybenko, Sufficiency of a Single Hidden Layer Feed Forward Network to Approximate any Continuous Function (Technical Report), University of Illinois, 1988, pp. 2-3.

Jiawei Han, Micheline Kamber, Data Mining: Concepts and Techniques, Harcourt India, San Diego, CA 92101 – 4495, 2001, pp. 5-6.

Simon Haykin, Neural Networks: A Comprehensive Foundation, Pearson Education, McMaster University, Hamilton, Ontario, Canada, 2004, pp. 2-4.


  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution 3.0 License.