An inductive learning approach to market segmentation based on customer profile attributes

Abstract

Inductive learning is a way of learning by showing examples to a learner (human or computer). The learning result is a conclusion drawn from the training data provided in the examples. An inductive learning approach is proposed in this study to assist in classifying customers into a number of market segments. In a particular segment, customers share similar needs and wants. The learning examples consist of both the positive training set (i.e., examples relevant to the learning theme) and the negative training set (i.e., examples irrelevant to the learning theme). Based on customer profiles and customers’ responses to previous marketing events, relationships between customer attributes and their preferences to marketing activities are constructed. Such a learning result can be applied to an unseen customer profile to determine whether the corresponding customer should be targeted for a particular marketing event. For further improving the learning result and thus, the overall segmentation performance, a learning feedback technique is proposed in this study to overcome the common drawback of inductive learning (i.e., incomplete training examples leading to an inappropriate conclusion). In an experiment, 1,500 anonymous customer profiles and customers’ responses to the marketing events were collected from a company. Among the 1,500 customer profiles in the entire collection, 1,000 of them were used as a training set while the remaining 500 customer profiles were used as an evaluation set. The results showed that 91.73% of relevant customer profiles were segmented correctly while 6.36% of irrelevant customer profiles were segmented incorrectly.