Extracting Potential Innovators Using Innovator Scores and Naive Bayes Classifier

Abstract

It is important to effectively penetrate a market with products in marketing of convenience products such as foods with short life cycles. It can be said that as products penetrate innovators, they will also penetrate late market segment. This study proposes a method to extract whether users are potential innovators or not based on the users' text data. We used users' innovator scores as the correct answer data and the morphological analysis of the text as the feature data, and applied a naive bayes classifier to extract innovators. As a result of validation using a beverage brand as the application object, we confirmed that a proposed method can extract innovators with high accuracy in method of classifying users into innovators and imitators. A proposed method is characterized by its ability to extract potential innovators on a large scale using a single free-form question item that is easy to input. In the future, we plan to improve the accuracy of the five cluster classifications of innovators, early adopters, early majority, late majority, and laggards, which are more effective for marketing analysis, and to apply a proposed method to many products to continuously verify the generality of a method.