Energy Analytics (Fall 2015)

I am teaching Energy Analytics, a graduate level course at UNC Charlotte, the second time this semester.

Student Profile

The class started with 8 master and 3 PhD students from various programs including master program in Engineering Management (3), master program in Electrical Engineering (4), PhD program in Mechanical Engineering (1), PhD program in Infrastructure and Environmental Systems PhD program (1), PhD program in Computer and Information Systems (1), and master program in Computer Science (1). Two PhD students (one from ME, and one from CIS) withdrew the course after the first mid-term exam. Since all students were on campus, the course was taught on campus.

Topics

The following topics were discussed throughout the semester with each topic covered in a week.

1. Greetings; Introduction; Naïve Models
2. Forecasting principles and practices 
3. Short term electric load forecasting
4. Long term electric load forecasting
5. Weather station selection
6. Probabilistic load forecasting
7. Electricity price forecasting
8. Wind and solar power forecasting
9. Forecasting and trading in electricity retail market
10. Demand response analytics
11. Outage analytics

Assignments and Exams

There were two in-class competitions, one on short term load forecasting, and the other on long term probabilistic load forecasting. Both competitions were also open to external teams. The short term load forecasting competition was counted as 20 credits, all based on forecast accuracy. The long term probabilistic load forecasting competition as counted as 60 credits, 40 credits on forecast accuracy and 20 credits on final reports and presentation. 

After each in-class lecture, I assigned a mini project, which is relevant to the topic of the week. The projects included forecasting stock price, implementing the methodology of a journal paper, presenting the contents of a notable journal paper, and so forth. These mini projects were counted as 10 credits. 

I also had the students to take some SAS courses, which were counted as 10 credits. If a student passes the SAS Advanced Programmer Certification Exam, I would offer 10 bonus credits. 

Training Objectives

The first offering of this course was organized around GEFCom2014, so this is in fact the first regular offering. When designing this course, I would like to train the students from the following aspects:

1. Reaching their potential. I made a commitment to the students "the more time you spend, the more you will learn". The in-class competition setup is to have them keep learning and trying new methodologies. 
2. Training their self-learning skills. I made them learn the software tools all by themselves without giving a single in-class demo. I also asked the students to read journal papers and reproduce the models. 
3. Practicing their presentation skills. In most of the in-class lectures, I crafted about half of the lecture time for the students to present their methodologies for the previous assignment. The presentation format includes both power-point and white-board. 
4. Mastering the documentation skills. In addition to the reports for the mini projects, the students were asked to prepare a substantial report for the probabilistic load forecasting competition. 
5. Understanding their competency in the field. I opened the in-class competitions to external teams, so that our students can see their rankings among the broader community. 

To-do List

Based on my observations so far, the objectives have been met. Nevertheless, there is still room for improvement:

1. Two of 11 students withdrew the class after the mid-term. Although the drop-out rate was better than the first offering (3 of 13 students withdrew the class), I hope to find a better way to have the students understand whether this is the right course for them.
2. Although the reading materials for this course were fairly comprehensive, a textbook would help further smooth the students' learning experience. I should complete my load forecasting book before the next offering of this course.
3. While the in-class competitions were open to external teams, I did not spend any effort validating their methodologies. Next time I should better reinforce the rules and let the teams publish their methodologies. 

It was a great pleasure for me to teach this course. I welcome any comments and suggestions.

Look forward to teaching the next class!