Learning to Imitate with Less: Efficient Individual Behavior Modeling in Chess