Research on User Segmentation based on RFL Model and K-means Clustering Algorithm

With the rapidly shifting dynamics of the current market, the companies are seeking a more thorough method to research the preferences of their target market. As such, data mining models of user segmentation are often utilized to fill up the broadcasting and television research areas. This paper proposes a broadcasting and television RFL model for channel user segmentation and then gives the model for typical use case. The model has two main advantages, showing the users' value dynamically and having strong data availability together with wide model applicability. To define users' degree of satisfaction towards diverse television channel, R, F and L indicators are built. Then this paper uses the optimized k-means algorithm to divide users into clusters, along with cross validation by two-step clustering, which helps verify the results. By comparing each user cluster's average R, F and L indicators with the ensample mean, users can be subdivided into six levels: key-growth user, key-development user, general-growth user, key-kept user, low-value user and general user. On this basis, recommendations are given to the broadcasting and television operators and advertisers to assist them to make profits.