ABSTRACT
Attributes of more than 7,000 customers of a product for shipping packages were analyzed to develop an automated system for predicting new customers. Critical parameters associated with these customers were identified by principal component analysis. Based upon these parameters and inter-relationships between them, a prediction model was developed to prioritize customer lists for the sales force in the field. An Artificial Neural Network (ANN) was developed to learn and characterize the non-linear model with an output that reflects the likelihood of a product sale in a percentile form. The ANN model was integrated in a user-friendly software package that will be used by the sales force to rank potential new customers. The software was completed and delivered for pilot testing.
