Abstract:
Customer prospecting has a great importance in today’s competitive business world. Organizations need to retain their customers for profitable operations. Choosing right customers for Customer Retention Management process is a difficult task. Data about customers is available all the time but analyzing it takes time and hence some automated solution is required. In this project we provide a supervised learning solution for solving customer-prospecting problem.
Many Artificial Intelligence techniques are available that can learn from a few examples. We chose Back Propagation Neural Network (BPNN) and Genetic Algorithm (GA) as underlying learning technologies for our project. We have used Gradient Descent learning algorithm with-in BPNN to achieve high accuracy.
Since transactional history is the only data readily and easily available to organizations, we chose to work on customer transactional data for finding out prospects.
The system has been tested for all the basic operations. It is robust, reliable, user friendly and showed very good accuracy for test data. We have tried different combinations of learning model parameters and found out the best combination for our test data. With best combination the average accuracy of the system is 82 percent and with controlled variations rises to 89 percent.
