A Naïve Bayes approach to map customer requirements to product variants

A company develops product positioning strategy to make each product cover certain market segmentation and meet a group of customers’ requirements. In this sense, customer requirements can be mapped to a product variant. This paper addresses the issue of mapping customer requirements to existing product offerings. We treat the mapping task as a classification problem. Product variants are used as the class label for customer requirements. Considering that customer requirements are usually expressed in ambiguous language and contain uncertain information, a probabilistic Naïve Bayes based classifier is built by using existing customer choices data. The classifier takes new customer requirements as input and the output is the product variant which the customer may be satisfied with. In addition, the probabilistic classifier leverages on the flexibility of customer requirements and classifies the requirements based on the probability of relevance of each product variant. Case study shows that the approach can achieve good performance in terms of classification accuracy and F-measure. Copyright © 2013 Springer Science+Business Media New York.