Evaluating TV Viewer Loyalty Based on TV Program Viewing History

Abstract

In existing researches, the methods of measuring viewer loyalties for TV program were limited to grasping the viewing situation of the program at one point of the program unit. These methods cannot evaluate based on viewer tendency, viewing continuity, or viewing time by viewer. In this research, we propose indices to quantitatively evaluate viewer loyalty for TV programs. By using these indices, it becomes possible to evaluate TV stations and TV programs from the viewpoint of viewer loyalty. In this paper, we apply these indices to a real TV program viewing history. We report several results of comparing and analyzing viewers classified by each indicator for each station and viewer segment, and predicting loyal viewers by random forest.