Machine Learning Approach to Day-ahead Scheduling for Multi-period Energy Markets