Neural Reinforcement Learning 1: Prediction