Learning Personalized Models of Human Behavior in Chess