Combining Reinforcement Learning and Inverse Reinforcement Learning for Asset Allocation Recommendations